Work in the Department of Zoology at the TUM in Munich

My call to the new Chair in Zoology at the Technische Universität München was in 1980. Our success in setting up research was aided initially by cooperations within the Sonder¬forschungsbereich (Cooperative Research Centre) 50 "Kybernetik" (Cybernetics) that was active in the Munich area. Within a couple of years, this was replaced by a new constellation of research groups in hearing sciences, Zwicker's Institute being supplemented by my own group and the arrival of Neuweiler and his people (Research in bat hearing and behaviour) at Munich's second University, the LMU. We formed the core of - initially - about 14 research groups, all working in hearing sciences in different institutions. The SFB "Hearing" was from the beginning a very active and coherent "group of groups" and was very successful in attracting support (a big thank you to the DFG) but also in producing a huge quantity of high-quality research in many areas of hearing. The financial flexibility permitted cooperation beyond Munich's borders, across Germany and the scientific world. One of my major emphases in this regard was the study of lizard hearing in one of their true homes, Australia.

Early research on otoacoustic emissions

At the time of the founding of the Zoological Institute of the T.U. München, the newly-discovered "cochlear echos", first described by David Kemp and later known as otoacoustic emissions, were starting to be intensively studied. I had, as one of my first contributions in this area (Manley, 1983c), suggested that, in mammals, the structure of the organ of Corti was such that it might contribute to the fact that most spontaneous emissions (that at that time were only known from lower frequencies in humans) would be fairly regularly spaced on the frequency axis. My reasoning was that the cochlear apex, where the lower frequencies are located, is tightly coiled. In order to maintain the cellular organisation of the hair-cell rows, hair cells in the inner and outer rows must either be of different widths (which they are not) or extra hair cells must be added at regular space intervals to the outer rows. This would, however, change the impedance pattern at that location, perhaps causing loss of acoustic energy to the fluid and thus an emission. This idea was later partially supported by work on primates, which also have a fairly irregular organ of Corti (unlike the rodents that are often used in hearing research).

Zwicker and I also carried out a study of emissions in guinea pigs (Zwicker and Manley, 1981), but failed to find spontaneous emissions. We resorted instead to stimulated emissions and were able to show that they were suppressed by low-frequency tones in phase-dependent way and in a similar pattern to those of masking-period patterns in human psychophysical data. This work greatly strengthened the impression that otoacoustic emissions were a very interesting and useful phenomenon, especially as a way of "remote sensing" what was going on in the inner ear. Such non-invasive techniques were much sought after by medical researchers for both clinical work but also for the possibility of direct comparisons between data from human and from animal studies.

We later carried out a great deal more research on otoacoustic emissions in birds and lizards and this work is described elsewhere on this web site.

Research on papillar anatomy and electrophysiology: Birds

Research on avian hearing continued apace, with a large study of the characteristics of starling auditory-nerve fibres being published (Manley, 1985), together with some observations of a still-unresolved phenomenon: preferred intervals. When auditory nerve fibres produce action potentials in the absence of sound, we refer to this as spontaneous activity. As such, it is a common feature of sensory systems, helping to maintain great sensitivity in the response thresholds. In mammals, the activity varies stochastically and the resulting distribution of time intervals between spontaneous action-potential firings is Poisson-like, being modified at very short intervals that, because of the time course of the action potential itself, cannot exist. In birds - and later to some extent in lizards - we were able to show that often, the histogram of intervals was modified, such that particular intervals with a period similar to that of the best response frequency, occurred more often than expected. These we called "preferred" intervals (Manley and Gleich, 1984) and showed that they were not the result of acoustic stimulation, since the sound was turned off (or even de-coupled) and the thresholds of some of the fibres showing this phenomenon were too high to be responding to extraneous sounds. We now believe these intervals were manifestations of the spontaneous activity of the active processes now known to exist in all vertebrate hearing organs.

A further important data set was obtained during the diploma thesis studies of two students in 1987. Alexander Kaiser and Jutta Brix were given the task of mapping the frequency distribution along the chicken auditory papilla at two ages shortly after hatching (tasks virtually unthinkable under the time restrictions of modern study conditions). The reasoning behind my assigning this work lay in a theory that at that time was new and based mainly on sound-damage studies in young chickens. On the reasonable assumption that loud sounds damage the papilla where the frequencies are analysed, this study reported that the place of sound damage was different at different ages. From this grew a proposal articulated by Ed Rubel that during ontogeny, the frequency map of the chicken cochlea moves along the papilla. The high-frequency end was supposed to respond initially only to low frequencies, which later in life are found apically. I was troubled by these data and their interpretation and decided that, in order to study this complex issue, a technique using low-level sounds to map the frequencies was necessary. Jutta and Alex mapped the frequencies in chickens two days and 21 days after hatching and, comparing the maps, we found them to be statistically identical - there was no shift of frequencies along the papilla during development. This finding was of sufficient importance to find a place in the journal Science (Manley et al., 1987). Much later (Manley, 1996), and in response to continued support for the idea of a shifting frequency map during ontogeny (nice ideas die slowly), I reviewed all of the earlier work that formed the basis of or had supported this idea and found it possible to explain without proposing a frequency shift. As an example of one of problematic technical issues in Rubel's work, isolated very young chickens were exposed to loud sounds while caged in a sound-proof room. In my lab, Alex Kaiser showed that such young chickens are only able to maintain their body temperature if they can huddle in a large group. When alone or in small numbers, their body temperature falls steadily (Kaiser, A. (1992): The ontogeny of homeothermic regulation in post-hatching chicks: Its influence on the development of hearing. Comp. Biochem. Physiol. 103A, 105-111). Knowing that frequency tuning along the papilla is highly temperature sensitive in non-mammals (Smolders in Frankfurt had shown a huge sensitivity up to one octave per 10°C in the pigeon), this means that the place in the inner ear that responds best to the damaging sound shifts continuously during a long sound exposure - and shifts down, towards the cochlear apex. Thus it is not even possible to say where one would expect to find the place of the greatest damage along the papilla from very loud sounds of specific frequencies.

In addition, and with remarkable tenacity, Sherri and Tim Jones, then of Nebraska, now in South Carolina, mapped the frequencies in the cochleae of pre-hatching chickens and found that there too, the map of the frequencies that were already present matched those we had found in the post-hatching cochlea (e.g., Jones, S.M., Jones, T.A., 1995b, The tonotopic map of the embryonic chicken cochlea. Hear. Res. 82, 149-157.). High frequencies were not present in very young chickens, the high-frequency part of the cochlea, far from responding to low frequencies, was underdeveloped and indeed silent. The addition of their work was conclusive and, indeed, a technical masterpiece, since recording from soft chicken embryos is an enormous challenge.

In 1988, a further student of mine, Christine Köppl (and now the principle person of this web site!), who had studied lizard auditory-nerve fibres for her diploma and doctoral thesis (see below) joined me in a working visit to the laboratory of Masakasu (Mark) Konishi at Cal Tech in Pasadena, California. Mark had been my thesis supervisor in Princeton, N.J. and had set me off an my path studying "reptile" hearing while himself then adding work on the barn owl to his repertoire. As a nocturnal hunter, barn owls catch their prey by listening to their sounds and localizing them precisely. In Pasadena, Mark had continued these owl studies and we joined his lab briefly to carry out research on one auditory nucleus of the brain stem that was involved in the binaural processing of sounds. This nucleus, then known as VLVp, or the nucleus ventralis lemnisci lateralis pars posterior, lay about 16mm deep in the brain. Mark's stereotactic apparatus made it possible to select angles appropriate to making contact with cells in this very deep nucleus (verified by lesions at the electrode tip), but the tension was high at the start of each experiment. Mark often came by early to ask, in his inimitable way, "You hit?". Hit we did and were able to show that the cells of this nucleus were stimulated by input from the contralateral ear but inhibited - a very little later - by input from the ipsilateral ear,. The cells were systematically arranged in the nucleus, forming a map of sound-level differences between the ears from the top to the bottom of the nucleus (Manley et al., 1988). This nucleus feeds to higher centres in the brain, where its input is integrated with information of the sound azimuth and thus the VLVp plays a key role in the interaural-level-difference pathway and sound localisation.

We planned a broad study of bird hearing and, to assess which groups would be interesting to compare, we began a long series of studies of the anatomy of the papilla using light- and electron microscopy. In 1988, Christine Köppl, Otto Gleich and Franz Peter Fischer and I published papers on papillar anatomy in the barn owl, starling and pigeon (Gleich and Manley, 1988; Fischer et al., 1988). One particular focus of interest was on the pattern of the orientation of the hair-cell stereovillar bundles, that is, the orientation of the axis of their sensitivity to sound-induced displacement. Unlike in mammals, avian hair cells are not arranged in strict rows, but in a mosaic. Hair-cell bundles in mammals all orient more-or-less parallel to the edge of the organ of Corti, whereas across and along the avian mosaic, the bundle orientations change dramatically. Especially in the apical (low frequency) half of the papilla, hair-cell orientation changes from parallel to the papillar edge on the neural side, to up to 90° turned towards the apex in the middle, and again parallel to the edge on the abneural side. We had no idea what this pattern meant functionally and, even today, the ideas that we do have are sketchy - but see below. In later years, further comparative-anatomical studies from our lab were published on the cochleae of canaries, zebra finches, budgerigars, emus, chickens and ducks (Fischer et al., 1992; Manley et al., 1993, 1996; Gleich et al., 1994; Köppl et al., 1998). In addition, Christine Köppl examined the auditory nerves of several bird species and quantified the sizes of afferent and efferent population axons (e.g., Köppl, C., Wegscheider, A., Gleich, O., Manley, G.A. (2000) A quantitative study of cochlear afferent axons in birds. Hearing Research 139: 123-143.), revealing interesting differences between species. This all made it possible to develop a comprehensive understanding of the structure and function of the bird cochlea (Manley 1995a; Manley et al., 1988, 1989) and of the phylogenetic development of the avian cochlea and its innervation (Manley and Köppl, 1998).

The rotation of the entire bundle of a centrally-lying bird hair cell raised the question as to the stimulation of such hair cells. If the papillar-tectorial complex vibrates across the radial axis of the papilla, it would optimally stimulate the hair cells on neural and abneural edges, which are oriented along this stimulation axis. But what of cells that are oriented up to 90° to that axis, "facing" apically? Classical work of Hudspeth's group on Frog sacculus hair cells had shown that if a stimulus is rotated 90° to the hair cell's optimal axis, the hair-cell response to simulation essentially drops to zero. This depends on the presence of molecular connectors between the tips of neighbouring stereovilli. These links are directly connected to the transduction channels, which are opened when the links are pulled upon by relative motion between stereovilli. These links are not pulled by a stimulus at 90° to the bundle's axis. In collaboration with Jim Pickles (then of Birmingham University, now in Brisbane, Australia), we studied the tip links of avian hair cells across the avian papilla, using the Scanning EM and tissue-fixation techniques known to preserve these links. The question was: are the links between stereovilli normally-placed in the bundles of centrally-lying hair cells or are they also rotated? The answer is simple: they are not rotated but situated normally (Pickles et al., 1989, 1990). This implies either that these cells are poorly stimulated by sound or that the pattern of stimulation is not simply radial across the papilla. Poorly-stimulated hair cells would result in auditory nerve fibres of higher threshold. At that time, we only knew that the spread of thresholds in recordings from any given bird was quite large (50 to 70 dB).

To further study this physiologically, my doctoral student Otto Gleich recorded from starling ganglion cells while attempting to stain individually-characterized fibres and trace these anatomically after the experiment. This is a challenging procedure and, to avoid ambiguity in stain identification, at most two or three staining attempts can be made per ear. Of course, not all of these stains work. Thus a great deal of time and effort is necessary to collect data on even 10 fibres, and the comparison of position and threshold can only be carried out over a narrow range of frequencies. Otto was, however, successful in collecting information on 12 fibres within the frequency range 0.6 to 1.8 kHz. These fibres showed a strong correlation between threshold and position (see Gleich, O. (1989): Auditory primary afferents in the starling: Correlation of function and morphology. Hear. Res. 37, 255-267.) across the epithelium, such that cells near the centre of the epithelium were 60dB less sensitive than those near the neural edge (n = 12, r = 0.764, P <0.01). Thus the large threshold range of avian auditory nerve fibres is traceable to the fact that these fibres innervate cells at different positions across the papilla and these hair cells are more and more rotated from the stimulatory axis. This can be seen as a specialization to obtain rate-level information from a large range of stimulus intensities.

Over the course of my tenure as Department Head of Zoology at the TUM, we had the great privilege of hosting four "Preisträger" of the Max-Planck Society. In 1989, Prof. Peter Narins from the University of California at Los Angeles worked with us. Peter and Otto studied the phase behaviour of starling auditory-nerve fibres (Gleich, O., Narins, P.M. (1988): The phase response of primary auditory afferents in a songbird (Sturnus vulgaris L.). Hear. Res. 32, 81-92¸ Narins, P.M., Gleich, O. (1986): Phase response of low-frequency cochlear ganglion cells in the starling. In: B.C.J. Moore, R.D. Patterson (Eds.), Auditory Frequency Selectivity, Plenum Publ Corp, New York, pp. 209-215 ). Several years later, a second MPS Preisträger, Prof. Bob Dooling of the University of Maryland, USA, also cooperated with Otto, working among other things on a special race of Canaries. These birds, so-called Waterschlager canaries, are popular with canary breeders because they sing very loudly. Otto and his partners were able to show that this was in fact due to the birds being partially deaf! (Gleich et al., 1994). Later work with this canary race (in further cooperation with Dooling, but also Georg Klump of my Department and other colleagues in later years) showed that these animals have genetic abnormalities that keep them in a constant state of losing and regenerating their sensory hair cells (Gleich et al., 1995). At any one time, a certain proportion of their hair cells are non-functional or dead. This not only explained their poor hearing but provided a highly interesting model for the study of structural and physiological regeneration of hair cells. It turned out that birds can always regenerate lost hair cells, but mammals cannot.

We were curious that, with the exception of only two fibres at the extreme apical end of the papilla, Otto Gleich found no fibres traceable to the abneural half of the papilla. At the time, it was only known from anatomical studies that the afferent innervation of the abneural hair cells was weaker than that of the neural hair cells. At that time, these two groups were classified as short (SHC) and tall hair cells (THC), respectively and differentiated arbitrarily by the tall hair cells being taller than wide, the SHC vice-versa. It was the painstaking work of my late co-worker Franz Peter Fischer, using serial transmission EM sections of bird papillae, that finally clarified the picture concerning innervation and function. Franz Peter was involved in almost all the anatomical work on bird papillae that took place in my lab, even if only as an advisor. His personal metier was the TEM work and here, he made an astounding discovery. Franz Peter was able to show that for a large population of SHC, the afferent innervation was not just weak, it was in fact totally absent! (Fischer, F.P. 1994, General pattern and morphological specializations of the avian cochlea. Scanning Microsc. 8, 351-364). Afferent innervation is strongest on hair cells near the neural edge of the papilla and weakens up to roughly the middle, beyond which the hair cells only receive efferent innervation. This, of course, immediately explained why Otto - and in the meantime others - was unable to find stained afferents in the abneural area of avian papillae - there aren't any afferents there. Franz Peter made the very sensible proposal that, now that we have an objective criterion for defining short and tall hair cells, we should abandon the arbitrary differentiation based on cell shape and use the innervation difference to describe THC and SHC. This was the first report of a very significant proportion of sensory cells lacking an input connection to the brain and raised, of course, the question - what are these cells doing?

All of this work fell in a time when in mammals, too, the functions of hair-cell populations were being discussed. In the mammalian cochlea, the inner hair cells are massively innervated (contacting many more afferents than an avian hair cell) but outer hair cells are weakly innervated by afferents and, like avian SHC, strongly innervated by efferents. In several papers attempting to describe these patterns and their possible functional significance, we proposed that, in fact, avian and mammalian cochleae, although having evolved their size and hair-cell patterns independently, have essentially evolved in parallel (Manley and Köppl, 1998; Manley et al., 1998). We suggested that the selective forces working in both papillae as they grew larger and more complex over evolutionary time led to a similar specialization of hair-cell populations. In that time, avian SHC have lost their afferent innervation, while in mammals it has been severely weakened but not (yet?) fully abandoned. Since avian SHC have no input connections to the brain, if they have a function (and being hundreds and thousands, they surely do), that function must be confined to within the papilla. One obvious possibility would be for them to mechanically influence the activity of the THC via the tectorial membrane. In birds, most THC are not placed over the basilar membrane, but on solid limbic material, thus an influence between hair-cell groups via the basilar membrane is not possible.

In cooperation with Prof. Charles Steel of Stanford University (also a MPS Preisträger), a micromechanical model of the possible behaviour of the tectorial membrane was calculated and published (Steele, C., 1997, Three-dimensional mechanical modelling of the cochlea. In: Lewis et al., (eds) Diversity in Auditory Mechanics, Singapore, World Scientific, pp. 455-461). This model indicated that if SHC were able to transmit mechanical energy from their stereovillar bundles into the tectorial membrane, that energy would collect and have its greatest effect near the neural border of the papilla - exactly at the position where the hair cells and their afferent fibers have the greatest sensitivity! The input of mechanical energy from SHC was at that time eminently conceivable, since both in mammals and in birds there was evidence for active processes in hair cells and a number of indicators pointed to a specialization of the SHC of birds and the OHC of mammals as the motor generators of the cochlea. My doctoral student Jutta Brix used a very sensitive motion-detector system developed by our late colleague Graeme Yates of Perth for the light microscope. She studied isolated avian hair cells and isolated pieces of papillar epithelium for evidence of activity in response to micromechanical stimulation (tiny water jets) and electrical current. There was some evidence (Brix and Manley, 1994), but these experiments were very difficult, since the hair cells are more firmly embedded in the tissue than are those of mammals. The question of the localization of active processes in the avian cochlea is still not fully described to this day. During this period, we collected much data and discussed a great deal the structure and function of the avian cochlea and this resulted in publications on the specializations of vertebrate hair-cell populations and their evolution (Manley and Köppl, 1998; Manley et al., 1998).

The cochlea of most land vertebrates (excluding mammals) contains not only the hearing organ, but a lagenar macula. This sense organ strongly resembles vestibular maculae, but its function was controversially discussed. Some evidence derived from staining the eighth nerve and examining which brain regions receive stained fibres had suggested that the lagena innervated the cochlear nucleus and thus could be auditory in function. A diploma student working with Jutta Brix in my lab, Christiane Haeseler, studied this question by staining physiologically-characterised fibres in the ganglion that had unusual spontaneous behaviour and/or did not respond to sound stimuli. These fibres were all traced to the lagena where their innervation patterns correlated with their spontaneous activity patterns (Manley et al., 1991). Thus it became obvious that the lagena is a vestibular and not an auditory organ. My doctoral student Alexander Kaiser also showed in his thesis that if the cochlear nerve and the lagenar nerve (which mostly run alongside and even intermingled with each other) are carefully stained separately, the lagenar fibres do not, in fact, innervate the cochlear nucleus but instead innervate known vestibular nuclei (Kaiser and Manley, 1996).

During a further sabbatical leave spent in Perth, Australia, I studied the inner ear of a ratite, or paleognath bird, the emu, in close collaboration with my postdoc Christine Köppl and the late Graeme Yates of the Perth lab. As adult Emu are somewhat unwieldy to handle in the lab (!), we studied the hearing of young emu chicks obtained from one of the many breeding farms that had been set up in the Perth region in those days. In the end, this turned into an in-depth study of emu cochlear physiology and anatomy and proved to be valuable for understanding the patterns we had been observing in our studies of the cochleae of birds from different, more derived, groups. Emus belong to the most basal group of birds and we therefore expected their cochleae to reflect this, providing information on the evolution of hair-cell specializations (Manley et al., 1997). Emu have especially sensitive low-frequency hearing, and we were able to observe its use in "grunt" and "growl"-type vocalizations in their social interactions on the emu farm. The frequency map of the cochlea is unremarkable, certainly of basal character, in being purely logarithmic (Köppl and Manley, 1997).

Emu hair cells also show basal features, such as being unusually tall. The cochlea containes a low proportion of SHC, one of the structural features we studied with Otto Gleich (who himself was carrying out a postdoc in the Perth lab at the same time) and with two of our highly supportive technical assistants in Munich, Gaby Schwabedissen and Elke Siegl (Köppl et al., 1998). One important feature of the discharge patterns of the auditory nerve fibres in the emu (and in a parallel study in the barn owl) was studied by Christine with Graeme Yates. Graeme had earlier developed a useful model of the rate-level functions of mammalian auditory-nerve fibres that showed that they all had a common breakpoint. This he attributed to the dependence of all mammalian afferents (that only innervate inner hair cells, the few OHC afferents are silent) on the summed mechanical activity of their region of the hearing organ. Christine and Graeme, using a large data base, discovered that in emus and barn owls, this is not true (Yates et al., 2000). Instead, different fibres have different break points, suggesting, importantly, and as was suspected from the anatomy, that their stimulation is a more local phenomenon and that the macromechanics of avian papillae is not as integrated as in mammalian organs. This, of course, is partly due to the placement of avian THC mostly on solid, limbic material and not over the basilar membrane. In mammals, the basilar membrane (BM) is thin and can absorb energy from OHC, forming a travelling wave reflecting the sum of all inputs. Thus BM movements in mammals reflect exactly what the hair cells are doing, in birds and lizards this is not the case at all.

An exciting discovery in the barn owl cochlea was a highlight of our avian work. My then postdoc Christine Köppl set up a colony of barn owls (Tyto alba) in the Department. Since barn owls are relatively rare and protected, we only studied owls that were hatched in our own colony. In cooperation with raptor societies and the nature authorities, quite a number of excess birds were released in a controlled fashion into the wild. Christine recorded from the barn owl nerve and brain and stained many physiologically characterized afferent nerve fibres to the cochlea and traced them to their innervation sites. She discovered that the frequency map of this fascinating species, far from being logarithmically arranged along the epithelium (as in "normal" bird and mammal cochleae), was strongly distorted (Köppl et al., 1993). The highest octave of hearing, which in the barn owl is exceptionally high for a bird species, from 5 to 10 kHz, occupied the entire basal half of the very long (11mm) papilla. Such a greatly "stretched" frequency representation had previously been found only in extremely specialized bat cochleae. Taking the designation used in bats, we called this an auditory "fovea", in the style of the foveal region of the eye. Behavioural evidence from Mark Konishi's lab had already shown that this is the frequency range that barn owl's use most to capture their prey. Thus here was evidence of an enormous evolutionary influence of life style on the structure and function of a major sense organ.

This was the beginning of a number of very interesting studies of barn owl ears, brains and hearing. Christine Köppl studied the anatomy of the basilar papilla (Fischer et al., 1988) and auditory nerve (afferent and efferent components) and showed that despite the distortion of the cochlear map, the frequency selectivity of the auditory nerve fibres is sharp but relatively normal (for a bird; Köppl, C. (1997): Frequency tuning and spontaneous activity in the auditory nerve and cochlear nucleus magnocellularis of the barn owl Tyto alba. J. Neurophysiol. 77, 364-377). The auditory-nerve fibres of the barn owl have, however, an extraordinarily acute ability to phase lock to tonal frequencies well above any frequency known from other birds or from mammals (Köppl, C. (1997): Phase locking to high frequencies in the auditory nerve and cochlear nucleus magnocellularis of the barn owl, Tyto alba. J. Neurosci. 17, 3312-3321). It is certain that this phase-locking ability plays a vital role in the coding of time cues of stimuli that are vital for sound localisation. In a comparison with the responses of auditory neurons in the two cochlear nuclei of the barn owl, Christine was also able to show how the coding of sound stimuli differed between the nuclei and across the frequency range. We later hosted Prof. Catherine Carr (University of Maryland, USA), also as a Humboldt Preisträger, and Christine and Catherine worked together on the nuclei of the owl and the chicken brainstem to examine the question of the evolution of the sound-localisation pathways in birds (Köppl, C., Carr, C.E. (1997): Low-frequency pathway in the barn owl's auditory brainstem. J. Comp. Neurol. 378, 265-282).

Parallel to these studies, Alex Kaiser had continued to trace fibres from the cochlea and lagena to the brain, and the main portion of his doctoral thesis concerned the activity patterns and the localization of efferent fibres in the avian brain. He was able to show that the efferents recorded in brainstem nuclei are activated by sound stimuli and that there are different populations of efferent fibres (Kaiser and Manley, 1994). Efferents show long-latency, broadly-tuned responses to tones, with discharge patterns typically different from afferent fibres. Two excitatory and one suppressive response type was identified, more types than in mammals. Thus Alex showed that the avian efferent system shows sound-driven responses, is complex and that the efferent fibres originate in a ventral group of cells. Efferents to the vestibular system originate in a dorsal cell group, as Alex showed for lagenar efferent fibres (Kaiser and Manley, 1996).

With two exceptions, our work on avian hearing in Munich reduced after 2000, being only continued by Christine Köppl until she left for the University of Sydney in 2007. Further studies continued with the (up to that time parallel-running) examination of lizard ears and hearing. However, Christine was very active, producing studies (that are further described in her sections of this web site) on avian efferent neurons in the brain, on a detailed study of the development of anatomy and physiological characteristics of the inner ear of the barn owl. Additionally, she continued work with Catherine Carr, especially describing the coding of sound stimuli in the cochlear nucleus angularis and the coding of interaural time differences in the brain of the chicken.

The two exceptions are (a) a study in cooperation with the Max-Planck Institute for Ornithology on the hearing of the capercaillie (Wood Grouse). We aided in setting up an acoustic system and computer software (thanks here to Prof. Hugo Fastl) capable of generating infrasound frequencies (below 20Hz). In two studies, we showed that the capercaillie produces (through powerful flutter-jumps), but does not respond to, infrasound (Lieser et al., 2005, 2006). This was important with respect to possible interference with its breeding habits due to infrasound from large wind generators near their Leks. (b) The second exception was studies initiated by Otto Gleich, then having moved to Regensburg, on the hearing of Dinosaurs. This non-experimental (!) study was based on the clear evidence, e.g., from paleontology, that birds are closely related to - in fact are - dinosaurs. The correlations that we had found between the structure and the function of the cochlea of living birds made it possible to predict some aspects of the function of long extinct dinosaur species for which the cochlear dimensions were known. Especially the advances in CT scanners have made it possible to study structures, such as cochleae, embedded deep within fossil heads. Using the data from two extinct dinosaurs (one Tyrannosaur and one Brontosaur) and from the extinct bird Archaeopterix together with structure and function data from many extant birds, we were able to predict the hearing ranges of these extinct animals. Not unexpectedly, dinosaurs and, indeed, the early bird species, heard over a restricted, low-frequency range (Gleich et al., 2005). This idea was taken up by colleagues at the British Museum in London, who in cooperation with me published a similar study showing that the anatomy of extinct ears can be used as a proxy to study their auditory capabilities (Walsh et al., 2009).

Our birds were, of course, also studied - partly in parallel to the work described above - with respect to the phenomenon of otoacoustic emissions. This other work is described in the appropriate section of my report, "Work on otoacoustic emissions".

Research on papillar anatomy, electrophysiology and modelling: Lizards

After my transition to the Chair in Zoology at the T.U. München, it took a number of years to begin again with studies of lizard hearing. This work was kick-started, however, during research visits to the lab in Perth, Australia in 1985 and 1989. We were interested in establishing studies on one species of lizard that could be thoroughly examined in many aspects, so that we could begin to really understand it and - in cooperation with Graeme Yates in Perth - model how it works. There had been two earlier studies of the bobtail skinkTiliqua rugosa (previously called Trachydosaurus rugosus), known locally as the sleepy or pine-cone lizard. This is one of the largest skinks and is still very common and experimentally very robust. One of the earlier studies had claimed that the hearing of this species varies with the time of year and that they only hear well during their active months, around September-October. Our first set of experiments was a thorough study of the activity of auditory-nerve fibres, their spontaneous and sound-driven activity, their response to temperature and seasonal changes. The results were published in a block of five papers (Manley et al., 1990a,b; Köppl and Manley, 1990a,b; Köppl et al, 1990). We were surprised to find that, in our hands, the sensitivity of hearing showed no changes across the seasons. However, what did change was the animal's requirement for anaesthetics to reach the same level of anaesthesia. The previous authors had used the same dosage throughout the year, a dosage that had been set during their active time. This turned out to be too much during their inactive season and under those conditions, the ear was insensitive or silent due to an anaesthetic overdose which, however, did not kill these robust animals.

The ear of Tiliqua, as studied in detail by Christine Köppl (now the principal person of this web site), consists of about 2000 hair cells that, as in all lizards, belong to two types (Köppl, C., 1988, Morphology of the basilar papilla of the bobtail lizard Tiliqua rugosa. Hear. Res. 35, 209-228). One type is found in an oval area at the papillar apex and is covered by a very large "blob" of tectorial material known as the culmen. Cells in this area respond to low frequencies below about 0.8 kHz and their frequencies are, unusually, distributed across the papilla. Frequencies above this and up to about 4.8 kHz are represented in an adjacent, long, tonotopically-organized area extending to the papillar base (Köppl and Manley, 1990). These hair cells are covered by a chain-of-pearls type of tectorial structure, a chain of so-called sallets. The frequency-tuning curves are sensitive (at best only slightly above human thresholds) and relatively narrowly tuned and have a characteristic shape (Manley et al., 1990a). These fibres were traced to small areas of hair cells across a narrow strip of papilla that generally corresponded to the width of a single sallet. Below 0.8 kHz, tuning curves are U-shaped, above this they have sharp tips. The frequency borders are temperature sensitive, moving up with increasing warmth at a rate of about 0.03 octaves/°C. The frequency figures given above are those for 30°C. Remarkably, the phase delays measured for auditory-nerve fibres were essentially the same as measured in the much longer hearing organs of mammals, indicating that these delays arise - in the absence of a travelling wave - purely through the varying response times of the hair cells of different frequencies (Manley et al., 1990b). This suggests that in mammals, the travelling wave delays are also caused by hair-cell latencies - and not the other way round. That the wave "travels" may thus be regarded as an epiphenomenon.

I write here "in the absence of a travelling wave", since, in cooperation with Graeme Yates, we measured the frequency response of the basilar membrane (BM). It was possible to surgically expose the underside of the BM, place light-reflecting particles on it and measure their movements in response to tones of different frequencies and levels. This surgery also exposed the entire nerve sheet, where the auditory-nerve fibres left the papilla. We were also able - within a few minutes - to record from single nerve fibres at the edge of the BM precisely at the locations where we had measured BM movements. In this way, we were able to compare directly the tuning of the BM and of the nerve fibres for different cochlear locations (Manley et al., 1988; 1989). It turned out that the BM was only very broadly tuned, and at all locations essentially followed the tuning of the middle ear (which we had also measured). The tuning of the nerve fibres was, however, under the same physiological conditions, very much sharper. In comparing these findings, we were able to derive the tuning function performed by the hair-cell-sallet-complexes sitting on top of the BM (manley et al., 1988). In these lizards, the tuning is essentially performed by small groups of hair cells and their sallet. The result is then picked up by locally-restricted innervating nerve fibres.

These data showed clearly that, unlike in mammals, frequency tuning in lizards is a local process that only depends on small groups of hair cells. In comparing this to other data from lizards with different tectorial membranes (no tectorial membrane in the alligator lizard Gerrhonotus, the work of Tom Weiss' lab in Boston, my own earlier work on the monitor lizard Varanus), I was able to derive a deeper understanding of the function of lizard ears and of their evolution. This work was based on extensive data bases from the late Glen Wever (of Princeton University, whom I met during my own thesis work there) and the late Malcolm Miller (of UC San Fransisco, who was a good friend and superb anatomist) and on our own anatomical work on lizards. The main concept underlying suggestions as to how these papillae had evolved was that the earliest lizard papillae consisted of three hair-cell areas, one low-frequency area central in the papilla and two high-frequency areas flanking this. These high-frequency areas were tonotopically organized, such that their highest frequencies were identical and lay at the two ends of the papilla - they were mirror-images of each other. During the evolution of the various lizard families, one of the high-frequency areas was either modified or lost, reducing the redundancy of frequency representation. Since either area could be lost, however, this led to the prediction that in lizard families that had lost the basal high-frequency area (e.g. geckos), the tonotopic organization would be reversed - an idea we were to test in later years. These concepts were collected and published in a number of comparative review articles (Manley, 2000a,b, 2002; Manley and Köppl, 1998).

Research on the structure of the ear of the Tokay Gecko, for which we had earlier studied auditory-nerve responses (Ruth Anne Eatock in Canada) was carried out by Christine Köppl and Stefan Authier, a diploma student (Köppl, C., Authier, S. (1995): Quantitative anatomical basis for a model of micromechanical frequency tuning in the Tokay gecko, Gekko gecko. Hear. Res. 82, 14-25). They found that the anatomical features that define the frequency response - e.g. the heights of the hair-cell bundles and the mass of the tectorial structures in Gekko gecko predicted also that the high-frequency area lay apical on the papilla and its frequency map would be reversed as compared to all other families of lizards and, in fact, to all other amniotes. We invited Mike Sneary, from California State University, San Hosé, to join us in a research project to map the frequencies in the Gekko papilla. This involved, as similar previous studies, the staining of single, characterized auditory-nerve fibres and looking for them in anatomical preparations of the papillae - which involves a lot of work. We made two interesting discoveries: (a) the tonotopic organization of the papilla was indeed reversed, (b) all the stained fibres innervated hair cells on the outer half of the papilla in the so-called postaxial area whose hair cells are covered by narrow sallets (Manley et al., 1999). The entire preaxial hair-cell strip was by-passed by the stained nerve fibres. This second fact remained a mystery until in 2005, Jim Hudspeth's lab showed using immune staining techniques that the reason was simple - there were no afferent fibres innervating the preaxial cells! We have since confirmed this result for the pygopod sub-family of geckos.

The Australian geckos are a very interesting group, since they are sometimes very different morphologically. There exists, for example, the subfamily Pygopodidae, the flapped-footed lizards. These are snake-like in form and have reduced their limbs to such an extent that only a tiny stub (flap) of the hid limbs remains. Remarkably, their inner ears strongly resemble those of other geckos and on these grounds alone they are obviously closely related. Pygopods are not too easy to find and are poorly known, thus they are protected by law in Australia. We obtained permission to carry out non-invasive field studies of pygopods in Western Australia, where there are the most species. For these studies, I chose a large area of the central Pilbara region (about 1600 km north of Perth) within which 7 species overlap. We were permitted to work in National Parks and on the huge Cattle Stations of the region. In order to be able to make measurements in the field, and thus avoid the necessity of repeated long trips from the field to the lab (unnecessary stress for man, lizard and vehicle!), I purchased a small caravan that could be used both as living quarters (extendable to two rooms) and laboratory. This work involved searching for pygopods, identifying them and testing their hearing in the field. Working in remote bush areas in Australia has a number of dangers (deadly poisonous snakes and spiders, among other thrills) and cannot be safely done alone. I was fortunate to win the cooperation of a diploma student, Hanna Kraus (now a doctoral student at Vienna University), who turned out to be utterly fearless and fast when catching lizards. She is also a consummate photographer, so that all animals could be photographically documented and the trips also brought many wonderful photographs of the areas where we worked.

Over the course of several, three-month trips between semesters, Hanna and I were able to collect specimens of several species of pygopods, both basal genera such as Pygopus and Lialis and several species of the advanced genus Delma. Hearing tests forming Hanna's diploma thesis consisted of inserting a tiny coiled-wire electrode into the mouth of a lightly gas-anaesthetized animal until the wire contacted the bone surrounding the inner ear (lizards have no secondary palate). All the equipment used was battery-operated, so that the work was under ideal conditions for noise-free recordings and, in the silent bush, was without sound shielding. Using tone pulses generated by a laptop computer, Hanna measured through the wire the summed electrical response of the auditory nerve, the compound action potential (CAP), to different frequencies and levels. The CAP "audiograms" of the basal genera were very much like those known from other geckos such as Gekko gecko, being centred near 1 to 2 kHz and with 6-7 kHz as their upper limit (at 75 dB SPL). Thus no surprises there. The CAP audiograms of the Delma species (four species were studied), however, showed a remarkable extra peak of sensitivity near 11 kHz, with an upper response limit near 14 kHz (Manley and Kraus, 2010). We also recorded vocalizations the animals made and their analysis revealed that such high frequencies are prominent in the sounds made by Delma, but less so or not at all in the vocalizations of Lialis - a species that is one of their main predators. It was exciting to be able to carry out such work totally isolated from human-induced noise and electrical disturbances and be able to return the animals quickly to the place of capture.

During other research visits in Perth, Christine Köppl and I also studied otoacoustic emissions in the bobtail skink extensively. These data are described elsewhere on this web site.

Research work on otoacoustic emissions of birds and lizards.

It is difficult to overemphasize the importance of the discovery, in 1978, of otoacoustic emissions (OAE) by David Kemp. These phenomena have had two very different areas of impact. First, they established a method (later developed into a series of methods) for studying the ear that are non-invasive and applicable to animals and man. Thus clinical work has enormously profited from this, as has our ability to compare hearing theory across the world of animals. Second, they led to a completely new understanding of how the ear works. Up until that time, the reason why auditory nerve fibres could be so sensitive, so sharply tuned but also labile had been the subject of wide-ranging speculation for decades. A short history of relevant research work and theories before the discovery of OAE can be found here (Cooper et al., 2008). OAE were a manifestation of sound energy being generated within the ear itself - hair cells could now be regarded not only as sound detectors but as independent generators of sounds. Their activity explained the exquisite sensitivity of ears. Two possible mechanisms for generating sound energy by hair cells were quickly discovered by Bob Fettiplace (then Cambridge University) and Bill Brownell (then Johns Hopkins University). In work on turtle hair cells, Fettiplace showed that the hair-cell bundles were not entirely passive structures. This led to much further work, especially in Jim Hudspeth's lab. Jim has also written a number of excellent reviews on the mechanisms involved. The involvement of transduction channels and tip-link tethers in generating active bundle movements is obviously a very ancient mechanism initially evolved in lateral-line and vestibular hair cells to overcome the viscosity of the fluid surrounding the bundles. Still controversial is the issue as to what extent active hair bundles still play an important role in mammalian cochleae.

The active mechanism recognized in mammalian cochlea and discovered by Brownell is the ability of outer hair cells (OHC) to actively change their length in response to a change of their cellular potential Since a response to sound produces a change in cell potential, the response itself can drive motor activity. This ability, which was not seen in mammal inner hair cells, turned out to be dependent on the presence of a protein (later called Prestin) present in densely-packed areas of the OHC's lateral membrane. In cooperation with Andy Forge (London University), Christine and I were able to show that the hair-cell membranes of geckos and barn owls (that we had shown produced OAE) did not contain similar densities of appropriate protein sizes (Köppl et al., 2003, 2004). This was evidence that in non-mammals, this second mechanism of generating OAE did not exist and thus non-mammalian OAE must arise from active bundle movements. However until the year 2001, such active bundle movements had only been observed in hair cells in vitro.

In cooperation with Graeme Yates and Des Kirk in Perth, we developed a method of examining the question of active bundle movements in an in-vivo preparation. Graeme and Des had been studying OAE emitted from the guinea pig while driving the auditory organ with electrical current in the inner ear and at various frequencies. Under these conditions, they could measure OAE sounds in the ear canal of up to 30 dB SPL. It was supposed that these sounds were generated by the electrical current driving the Prestin of the OHC. It occurred to us that the ear of the bobtail lizard offered a golden opportunity to test in vivo whether lizard OAE were generated by a Prestin-like mechanism or by the hair-cell bundles. Our reasoning was as follows: Unlike in mammals, lizard hearing organs contain hair cells populations whose bundles are oppositely-oriented. If electrical current would stimulate lizard hair cells, a prestin mechanism in oppositely-oriented hair cells would produce phase-identical OAE activity that would sum before being emitted. On the other hand if the active mechanism was in the hair-cell bundles, the activity produced by oppositely-oriented bundles would be of opposing phase and in fact interfere with each other before the resultant was emitted from the ear. Thus we adapted the current-injection technique to lizard ears and tested whether they even emitted measureable OAE under these conditions. They did, but the OAE were exceptionally small, leading us to suspect that we were getting subtractive effects in the papilla. To test this, we used very low-frequency sounds to bias the position of the hair-cell bundles during the injection of much higher-frequency current. We reasoned that when the low-frequency sound biased the bundles heavily to one side, this would be suppressive for bundles of one orientation and not suppressive for the other set of hair-cell bundles. During each extreme of papillar movement, one set of hair cells would emit no OAE. Thus the emitted OAE should have opposite phase for each half-cycle of the low-frequency sound. We found exactly that. Under conditions of added sound, the OAE were larger (the suppressive effects were on average smaller) and the phase of the OAE flipped 180° for every half cycle of the low-frequency tone.

This data was the strongest evidence yet that in non-mammals in vivo, OAE were the result not of hair-cell membrane-bound Prestin activity, but the result of active bundles. The resulting paper in PNAS (Manley et al., 2001) was one for which we were able to use most clearly the much overrated "hypothesis-driven" approach to research. Subsequently, I wrote two papers (Manley, 2000, 2001) whose explicit aim was to demonstrate that all the phenomena thought to be characteristic of an active mechanism in the cochlea of mammals were also evident in data from non-mammals such as birds and lizards. The comparative study of active mechanisms can not only help understand their evolution but also how such mechanisms function in all vertebrate ears.

Our first exposure to OAE work in non-mammals was earlier, during a research visit to the Perth lab, where Christine Köppl and I examined the bobtail lizard for such emissions. Unsure whether there would be anything resembling spontaneous OAE, we first studied distortion-product emissions (DPOAE). These arise when the internal distortion of the ear (produced by its own profound non-linearities) generates sidebands if two tones stimulate it. These so-called distortion products can be quadratic in nature (e.g. f2-f1, these being the two frequencies) or cubic, such as 2f1-f2 or 2f2-f1, etc.). As quite a lot was known about cubic DPOAE of humans and mammals, we first studied these. It turned out that bobtail lizard ears produced fairly strong DPOAE throughout the frequency range (Manley et al., 1993). Their complexity was, however, greater at low-frequencies. DPOAE showed rate-level functions resembling those of mammals (indicating among other things more than one location of the source) and they were suppressible by a third tone (Köppl and Manley, 1993). The suppression patterns were very interesting, since the tuning curves of the suppression phenomenon strongly resembled the single-fibre tuning curves we had measured in this same species. This was a very powerful indicator that the OAE phenomena derives directly from hair-cell activity that also produces a stimulation of afferent nerve fibres.

During the study of DPOAE, we noticed the presence of small spontaneous OAE (SOAE) peaks. These we later studied in great detail, also in many other lizard species (more than 30 in total), and they have become a very important source of information on the physiology of lizard auditory papillae. SOAE were found only in the high-frequency response region, below 1 kHz there were none (Köppl and Manley, 1993). The SOAE were generally fairly numerous, on average 10 per ear between about 1 and 5 kHz. They were broad-band and sensitive to temperature changes (Manley and Köppl, 1994), hypoxia (reduced lung ventilation) and the presence of sounds. Sounds could suppress, but also facilitate, SOAE levels and the suppression tuning curves again strongly resembled those derived from single auditory nerve fibres, indeed they were indistinguishable (Köppl and Manley, 1994). Thus we were very confident that the OAE truly were a direct reflection of the active processes in local groups of hair cells. The SOAE typically were found with a certain frequency separation, which we felt reflected the way in which hair-cell activity would be bundled by the coupling of hair cells within and perhaps between sallets. This suspicion was later confirmed in models of the papilla produced recently by two research groups for species with tectorial sallets, the bobtail lizard (Vilfan and Duke) and the Tokay gecko (Gelfand et al.).

Temperature changes and tonal suppression we later used in a variety of species having different papillar structures, particularly of different sizes and with different tectorial membranes. Thus we have published data on Anolis (an iguanid lizard that has no tectorial membrane, TM, in the high-frequency area) showing SOAE up to 7.7kHz and indications that single SOAE peaks can be generated by as few as two hair cells (Manley and Gallo, 1997). Comparative studies were also carried out in other species lacking a TM (Manley, 2006a), but also in skinks (Manley, 2009)and closely-related families and in geckos (Manley et al., 1996), in monitor lizards (Varanidae; Manley, 2004c)) and, currently in preparation, in teeid (Teeidae) lizards. While all skink-like families and geckos have (independently-evolved?) salletal tectorial material, varanid and teeid lizards have continuous TMs. Comparative studies had shown (a) that temperature effects are largest in species with a continuous TM and (b) in all species where the auditory-nerve single-fibre tuning is known, the suppression tuning of SOAE resembles this tuning very closely (Manley, 1997). Why the temperature effect should be so large in species with a continuous TM is still largely a mystery.

Since we had demonstrated that emissions in the Bobtail lizard Tiliqua derived from activity in the hair-cell bundles, we decided to manipulate the ionic environment of the hair-cell bundles to compare the results to similar experiments carried out with hair cells in a dish (in vitro). Especially frog sacculus hair cells had been examined in this way, but such hair cell bundles only show very low frequency spontaneous oscillations, far below those of lizard SOAE. In cooperation with the late Des Kirk of the University of Western Australia in Perth, I found that the frequency of SOAE could be manipulated by manipulating the level of calcium ions in the scala media. We first developed a suitable surgical approach for this technique and found that we were able to penetrate scala media with a large (> 6µm tip) electrode without disturbing the generation of SOAE. Positive and negative currents affected SOAE frequencies systematically (Manley and Kirk, 2001), a phenomenon which we examined thoroughly in order to separate its effects from current injections in which calcium was manipulated. We tried two approaches. First, we current-injected BAPTA, a powerful calcium chelator that rapidly produces a reduction of free calcium levels in the endolymph of scala media (Manley and Kirk, 2005). Not unexpectedly, BAPTA produced a rapid lowering of SOAE frequencies (of up to 25%), an effect well documented for hair cells in vitro. In cooperation with Des Kirk, Christine Köppl and Ulrike Sienknecht (also from the Munich lab) I studied the effects of injecting calcium directly. For this, we chose electrodes large enough (> 6µm tips) to passively and slowly "flood" scala media, rather than use current. The replacement of the endolymph in this way led to correlated changes in SOAE frequencies. An increase in calcium levels led to a rise in frequencies, albeit in a limited fashion, since the SOAE amplitudes then tended to fall quickly into the noise. A reduction in calcium levels led to a prolonged fall in SOAE frequency, depending on the calcium concentration (Manley et al., 2004). No effect of SOAE frequencies was measured for a concentration of approximately 1mM, suggesting that this is the normal level in endolymph. On the one hand, this was surprising, since in frog sacculus, earlier work had suggested the use of 100mM calcium levels as representing normal conditions. In mammals, calcium concentrations are extremely small (20 µm). These data can be understood if we remember that in non-mammals, the basilar papilla lies partly beneath but very close to the lagenar macular. This vestibular organ has embedded otoliths and it has been shown in fish that otoliths are in constant calcium exchange with the endolymph. Unless the calcium levels in endolymph are controlled, the otoliths can be chemically eroded. In mammals, the lagena has been lost during evolution, allowing a dramatic fall in calcium levels that no doubt have important (but as yet poorly understood) consequences for mechanical transduction channel activity. We suggest that in lizards, the endolymph has a calcium concentration near 1.0mM. These data may thus be interpreted to mean that in-vivo SOAE activity in lizards is the result of spontaneous mechanical activity of the ion channels of the stereovilli of the hair-cell bundles. Thus data from earlier in vitro studies of frog sacculus and turtle basilar papilla hair cells have a direct bearing on lizard SOAE and our general understanding of how hearing works.

In cooperation with Pim van Dijk (University Hospital, Groningen, The Netherlands), we carried out two studies on lizard SOAE. In the first (van Dijk et al., 1996), we showed that stronger lizard SOAE (and, indeed, barn owl SOAE) show statistical properties of their amplitudes that are characteristic of an active sound source and not those of a filtered noise. This supported the conclusion reached on other grounds (summarized in Manley, 2000 and 2001) that in non-mammals, also, there exists a cochlear amplifier (see also data discussed above). In the second study (van Dijk et al., 1998), we looked at the amplitudes of different SOAE peaks in the same ears simultaneously and showed that they interact with one another such that a change in amplitude of an emission may very rapidly be followed by a change in amplitude in another emission. These interactions were very fast indeed, with a mean delay time of only 0.2 ms, compared to several ms in human SOAE. In spite of this difference to humans, the actual frequency distances between interacting emissions was almost identical to that found in humans - despite the huge difference in the physical distances between emitting papillar areas (e.g. 0.5 mm compared to 6mm). This data added to the long list of similarities between the characteristics of lizard and of mammalian SOAE.

Since we were initially unable to detect SOAE in birds (chickens, starlings), we resorted, as in the guinea pig, to studying emissions resulting from tonal stimuli. Using low-level swept tones, we found that where emissions were present, the measured sound pressure of the tone in the ear canal varied with frequency and that these dips and peaks were largest at the lowest tonal levels used (Manley et al., 1987). This indicated that the low-level tones were interfering with emissions in a phase-dependent manner. The emissions were suppressible using additional tones and they proved to be suppressed in a highly frequency-selective fashion.

In chicken and starling, we later studied the basic characteristics of the distortion-product OAE (DPOAE) and measured 2fl-f2 and 2f2-f1 in the ear canal of both awake and anaesthetized starlings and chickens (Kettembeil et al., 1995). The effect of a third suppressive tone and the behaviour of the DPOAE under anaesthesia were also studied. In general, the DPOAE characteristics of both bird species resembled those of lizards and mammals, but first appeared at somewhat higher primary-tone levels. The best frequencies of third tones suppressing 2f1-f2 lay near the first primary tone (fl). Facilitation via a third tone was also seen, often at levels below those eliciting suppression. Remarkably, the DPOAE 2f1-f2 disappeared completely at the onset of deep anaesthesia and recovered to its original magnitude when the anaesthesia was lightened, sometimes, however, with a considerable delay. The compound action potential of the auditory nerve was, in comparison in the same ear, somewhat more sensitive to anaesthesia than the DPOAE. This suggests that the anaesthesia first affects the hair-cell synapse before the it shuts down normal transduction in the hair-cell stereovillar bundles. Control experiments showed that the anaesthesia effect was not an indirect result of a possible hypoxia. Avian DPOAE at low and intermediate sound levels are thus physiologically-sensitive manifestations of normal hair-cell function that are, in contrast to mammals, also anaesthesia-sensitive.

We later studied the barn owl with respect to OAE, work that formed the doctoral thesis of Grit Taschenberger, who is now in Göttingen. Grit showed that unlike the other avian species, barn owls do produce SOAE, each emitting ear producing on average 1.9 emissions (Taschenberger and Manley, 1997). Almost all of the emissions lay at high frequencies, within the acoustic fovea that Christine Köppl had discovered, between a frequency of 7.5 and up to 10.5 kHz and with peak sound-pressures between 35.8 and 10.3 dB. The SOAE were rather broad-band in frequency, but still narrower than lizard SOAE. The 3 dB bandwidths ranged between 4.5 and 11.4 Hz. As in lizards, the SOAE frequencies were temperature sensitive. Raising the temperature shifted the emissions to higher frequencies (by on average 0.039 oct/°C), and vice versa. External tones could suppress the level of SOAE, an effect that was highly tuned. For SOAE with frequencies between 2.5 and 10.5 kHz, the Q10dB values of 2 dB iso-suppression tuning curves (STC) varied from 1.07 to 10.40, values directly comparable to those Christine Köppl had measured in single auditory-nerve fibres. The best thresholds of 2 dB STC were generally below 15 dB SPL. Thus in this species, also, the OAE are useful tools for studying the function of the inner ear in a non-invasive fashion.

Grit also measured distortion-product otoacoustic emissions (DPOAE) 2f1-f2 in the ear canal of the barn owl (Taschenberger and Manley, 1998). DPOAE were elicited by primary tones in 11 frequency regions from 1 to 9 kHz. The highest DPOAE output levels and best thresholds were generally found for f1 frequencies of 4 to 7 kHz. DPOAE levels could be suppressed in a frequency-selective way by adding a third tone. As in other non-mammals, the best suppressive frequencies were near f1, suggesting DPOAE generation near the frequency place of this primary tone. This is in contrast to what is known for mammalian species, where the DPOAE is thought to be generated near f2. The Q10dB-tuning selectivity values of suppression tuning curves increased as a function of frequency up to 15.8. This tendency resembled the increase in frequency selectivity of SOAE suppression and of auditory nerve fibres in this species.

OAE in barn owls could be influenced by playing sounds into the contralateral ear and this effect is mediated via the efferent fibres to the hair cells. The avian auditory papilla provides an interesting object on which to study efferent influences, because whereas a significant population of hair cells in birds is not afferently innervated, all hair cells are efferently innervated (see above). Previous studies in mammals using contralateral sound to stimulate the efferent system had demonstrated a general suppressive effect on all otoacoustic emissions. Grit Taschenberger studied the effect of such stimuli (broadband noise, pure tones) on the amplitude of the DPOAE 2f1-f2 and on spontaneous otoacoustic emissions (SOAE) in the barn owl. For the DPOAE measurements, fixed primary-tone pairs were presented and the DPOAE measured in the presence and absence of continuous contralateral stimulation. The DPOAE often declined in amplitude but in some cases we observed DPOAE enhancement (Manley et al., 1999). The changes in amplitude were as large as 9 dB. The influence of the contralateral noise changed over time, however (putatively a result of fluctuating anaesthesia levels), and the effects of contralateral tones were frequency-dependent. SOAE were suppressed in amplitude and shifted in frequency by contralateral broadband noise. Control measurements in animals after middle-ear muscle resection (here we were grateful for the skills of Horst Oeckinghaus of our Department) showed that these phenomena were not attributable to the acoustic middle-ear reflex. The finding of DPOAE enhancement is interesting, because a type of efferent fiber that suppressed its discharge rate during stimulation was described in birds by Alex Kaiser in my lab (see elsewhere on this web site).

In a third cooperation with Pim van Dijk, the question was re-examined at to whether amphibian ears emit DPOAE. There was a statement in the scientific literature to the effect that they did not. We looked at this question in my Munich lab using lightly anaesthetized tree frogs of the species Hyla cinerea. The frog inner ear, of course, contains two hearing organs: the amphibian and the basilar papilla. In Hyla cinerea, the amphibian papilla is sensitive to low- and mid-frequency stimuli (0.1 to 0.5 and 0.5 to 1.3 kHz, respectively), whereas the basilar papilla is sensitive to high-frequency stimuli (2.8 to 3.9 kHz). We immediately found distortion product otoacoustic emissions (DPOAE) in each of six ears of six individuals investigated (van Dijk and Manley, 2001). The DPOAE at 2f1-f2 was present when the primary frequencies f1 and f2 and the DPOAE frequency were close to either the mid- or the high-frequency range. At frequencies between the sensitive ranges of both papillae, no emissions were observed. For the basilar papilla, the dependence of DPOAE level on the primary tone frequency ratio f2/f1 showed a pattern characteristic of the response of a single nonlinear resonator. Thus, in agreement with neural data from frogs, DPOAE from the basilar papilla reflect the contribution of a single auditory filter to emission generation.

The middle ears of amniote vertebrates, revisited.

In 1995 I wrote a short essay on middle-ear prostheses (Manley, 1995b). Medical workers had been concerned that the simple prostheses being used in medicine to replace damaged or ossified middle ears were not such a good idea. First, they thought that the columella structure might be pushed too deep by sound, an absurd idea when contemplating the incredibly tiny movements of the footplate (smaller than those of the basilar membrane!). They also thought that a three-ossicle middle ear might be better. However, the single ossicle ear is not inferior to the three ossicle middle ear at all and much easier to manufacture and insert.

Another middle-ear interest was awakened by new findings from work with fossil vertebrates. Palaeontological studies (e.g. by Jenny Clack at Cambridge University, England) had shown that for a long period in the early evolution of the amniote vertebrates, there was in fact a general absence of a tympanic middle ear. A tympanic middle ear is one with an eardrum or tympanum and a lever system to provide an impedance match between the fluids of the inner ear and the air outside. Within a short period in the Mesozoic geological era, all the then extant groups of amniotes (squamates, archosaurs such as birds, and mammals) developed a tympanic middle ear independently of each other. This amazing fact was the key to understanding why the inner-ear papillae of these groups are so characteristically different - they all enlarged and evolved independently. These facts we summarized in various publications and tried - with mixed success - to bring these findings into the consciousness of auditory researchers that lacked a zoological-paleontological background (Manley, 1973, 1986, 1988, 2000c, 2004b, 2010a, 2012a, 2012b; Manley and Clack, 2004; Manley and Köppl, 1988; Manley et al., 1986). The mammalian three-ossicle system works better at higher frequencies (there is no cartilage that can easily bend) and this fact favoured the evolution of high-frequency hearing in mammals. (Manley, 2010a), a controversial matter (Manley, 2010c, 2012b).

Having studied lizard middle ears in the early 1970s, I returned to them again in 2005 and 2007. Lizards have a potential problem localising sound, since they can only hear rather low frequencies and the head is too small to produce a useful sound shadow at those frequencies as the sound wavelengths are much greater than the head size. This deficit can be overcome through the use of a pressure-gradient receiver. Together with Dr. Jakob Christensen-Dalsgaard, I studied the middle ears of a number of lizard species during two research visits to Jakob's laboratory in Denmark. Previous to this research, work on insects, frogs and birds had shown that to various degrees and over various frequency ranges, animals whose middle ear is not essentially closed behind Eustachian tubes (as in mammals) but open to a cavity (such as the mouth) should show clear signs of interactions between the eardrums. Sound entering one ear will cross the head - more or less attenuated - and drive the eardrum of the opposite side. Thus, depending on the direction of the sound source, the response of the two eardrums will be more different than if there no connections within the head. Sounds from the contralateral side could interfere both destructively (reducing eardrum motion) or constructively (increasing eardrum motion) with sound from the ipsilateral side. Such interference can induce effective pressure differences between the ears that are much larger than the actual differences due to sound shadowing. Such a system is known as a pressure-gradient receiver, as opposed to the mammalian pressure receiver, and should provide the animal with clues on the direction of the sound source. It had been shown to work well in insects such as crickets, but the effects in frogs and birds defied clear definition and were much weaker than might have been expected. In frogs, it later turned out that the eardrums were not the only source of sound input to the mouth cavity, complicating the interactions. A nice, clean case of a functioning pressure-gradient received had not been demonstrated in land vertebrates. The aim was to examine whether the lizard middle ear is a pressure-gradient receiver or not. We used laser vibrometric technique to measure the lizard's eardrum velocities at different frequencies and under different conditions.

The research work showed that acoustical coupling of the two eardrums in lizards is very strong (especially in those with thin, delicate eardrums) and produces the largest directionality of any terrestrial vertebrate ear studied. This research resulted in two publications (Christensen-Dalsgaard and Manley, 2005, 2008). Laser studies of tympanic motion show pronounced directionality within a 1-2 kHz frequency band around the best frequency of hearing, caused by the interference of ipsi- and contralateral inputs. The results correspond qualitatively to the response of a simple middle ear model assuming coupling of the tympana through a central cavity. Furthermore, observed directional responses were markedly asymmetrical, with a steep gradient of up to 50-fold (34 dB) response differences between ipsi- and contralateral frontal angles. Thus lizards capture directionality information before any neural processing is necessary and the directionality is easily exploitable by simple binaural subtraction in the brain. Lizard ears are so far the clearest vertebrate examples of response directionality generated by tympanic coupling.

In mammals, by contrast, we are confronted with a pressure-receiver system. This system developed independently in early mammals (a caviat: there is evidence that such a system developed twice, once in monotreme prototherians such as the duck-billed platypus and once in therian mammals). Here, it is not yet clear whether (a) mammalian ancestors possessed an open connection between their middle ears, which later became very restricted, or (b) they developed their middle ear system completely de novo, without such a connection. Early mammals developed an eardrum in a deeper location than did non-mammals and they also developed a secondary palate lying between the middle ears. Due to the fact that mammals, in contrast to other amniotes, generally chew their food and begin digestion in the mouth, it is possible that in possibility (a), the mouth became more and more closed off from the middle ears. Together with an increasing brain size, this finally reduced the Eustachian-tube connections to narrow openings unusable for a pressure-gradient system. The evolutionary consequences of these events are great. The loss of a functioning pressure-gradient receiver (possibility a) meant that, in order to derive information on the direction of a sound source, the input to the two ears had to be compared by neurons in the brain. In that case, this could have been one reason for the expansion of the auditory brain and the evolution of refined auditory pathways comparing the discharges of neurones from the two ears to derive the exact level difference and arrival-time difference of sounds at the two ears. However, in the case that the middle ear cavities developed de novo, the evolution of brainstem analysis systems would have been equally pressing. The evolution of earflaps or pinnae also provided an enhanced directionality at the periphery. No birds or lizards have ear flaps, but all mammals do, even those that fly (and would find them an impediment to flight). As was noted above, it is also likely that the suitability of the mammalian middle-ear type to the transmission of high frequencies favoured the evolution of higher-frequency hearing and animals with small heads can better sound-shadow at higher frequencies. Thus high-frequency hearing presumably stood under great selective pressure in early mammals (that were all small) and was possibly enhanced in parallel with the possible loss of the pressure-gradient system. I recently held a plenary lecture held on this topic at the 2009 MEMRO conference at Stanford University, which has been published (Manley, 2010a). This has subsequently been further analysed (Manley, 2010c; Manley and Sienknecht, in press).