Rhinolophidae some Hipposideridae and one species of Mormopidae are the only mammal species known to have a “deformed” cochlea. Normally, mammals, like humans, have a cochlea that is logarithmic to some extent. The term logarithmic is very broad and can imply a base 2 or even a base 10 logarithm, which is a big difference! But what does it mean that our cochlea is “logarithmic”? The term logarithmic here refers to the frequency a mammal would listen to. If the frequency is low, say around 500 Hz, as components of the human voice are, even the tiniest change in frequency makes a different portion of our cochlea respond strongly. However, if the frequency is high, say a squeek of a bat you are just able to hear, a tiny frequency change would still stimulate virtually the same portion of your cochlea. To make the cochlear responses exactly as different as in the case with the low frequency, that squeaky bat really has to change its frequency a lot. So, from a cochlear perspective: low tones need tiny changes and high tones large changes to cause a shift in the cochlear portion that responds. If one plots the required frequency change against the frequency of the stimulation tone, we see that the relationship is logarithmic. It may not be quite coincidental that music instruments, such as a piano are based on an octave-system, which is in fact a logarithmic system with respect to frequency.
So all mammals have identical cochleas? No. They all have some kind of logarithmic scale, however, the one mammal is generally more sharply tuned over its frequency range than the other, but all seem to exhibit more or less the same need for larger tonal changes at higher frequencies, if this change is expressed as a fraction of the frequency of this tone. Humans have a hearing that is relatively sharply tuned around each tone, whereas rats and bats are not so sharply tuned, but we are all logarithmic: moles, mice, elephants, pipistrelles, except for? Horseshoe bats and some species of Hipposideridae. Rhinolophidae have a deformed cochlea. Their cochlear tuning is only precise around one particular frequency: the frequency of their second harmonic, which is the frequency used during hunting. Lower frequencies are less sharply tuned in the cochlea, which is quite an exception. Because of the stir this exception caused among scientists, most people now think that “bats” have a deformed cochlea, which is in fact only true for Rhinolophidae. Vespertilionidae, Molossidae, Megadermatidae and Phyllostomidae all seem to have dead normal cochleas for a mammal.
Individuals vs populations
This difference has to be kept in mind when comparing the use of frequencies in Vespertilionidae with Rhinolophidae. An individual vespertilionid bat is in principle free to choose whatever QCF frequency it likes, whereas a rhinolophid is constrained by its cochlear design. If a rhinolophid shifted its frequency only by 3 kHz, it would simply become impossible to echolocate. The only constraints vespertilionid bats have would be neural frequency-tuning constraints, which may even be flexible. Secondly, neural filters are not strongly tuned to frequency. Inviduals of a species of vespertilionid bat can therefore still echolocate while changing their QCF frequency. This is visible if the bat changes from long to short signals. Long, narrow band signals in Pipistrellus javanicus are at 39 kHz, but short signals typically end at 43 kHz. These shifts are typical of most vespertilionid bats. It has been discovered that horseshoe bats (Jones and Ransome, 1993) and Hipposideridae (Hiryu et al., 2006) can change their frequency over time as well, even as much as 0.43 kHz/day. It is yet unknown whether cochlear sensitivity will shift accordingly. However, the point made here is that Vespertilionidae regularly shift their QCF frequency by 4 kHz within one second whereas CF bats, like horseshoe bats use a fixed frequency during hunting. Each horseshoe bat has a fixed frequency, that may drift slowly over weeks and horseshoe bat populations, for example, bats from one cave, resemble each other strongly in CF frequency. Yoshino et al. (2006) found on the island Okinawa jima that Rhinolophus pumilus sharing a cave were more likely to use similar frequencies than individuals from elsewhere on the island. Geographical variation in calling frequency in Rhinolophidae and Hipposideridae is a big problem for identifying species of these groups in Indonesia.
No such phenomenon has ever been reported in Vespertilionidae. Pipistrellus pipistrellus is known to exhibit a large intra-specific variability in QCF frequency, but this can be recorded at a single location and no systematic geographic differences have so far been discovered in Europe. In the United States intraspecific variation also proved to be mainly due to individual / recording situation and no systematic geographical trends could be found in a number of vespertilionid bat species (Murray et al. 2001). An exception may be the genus Vespadelus in Australia, but this claim requires a more thorough investigation.
Consequences of frequency shifts on echolocation
A species of horseshoe bat calling at 80 kHz does not process its signals any differently from an individual, or even another species calling at 100 kHz. Using 80 kHz may give a slightly higher reach, but a slightly worse frequency resolution (also depends on many other factors), than using 100 kHz, but this difference is marginal. The reason for this is that horseshoe bats use a flutter detection system. They listen for frequency modulations in their echoes. The ability to detect such modulations is only marginally affected by such frequency differences. The frequency used by a horseshoe bat may therefore betray more about its ‘clan’, geographical location and species than allowing one to quantify its echolocation abilities. This situation is quite the opposite in Vespertilionidae. Long QCF calls at low frequencies will make a bat crash land very quickly in dense forest. Short, broadband calls at high frequencies could be used in open air, but the other bats flying in the same space would do much better in detecting insects over long distances. In the frequency range used by many Vespertilionidae for their QCF component of 18-50 kHz, wavelength changes dramatically, which impacts target strength, reflection patterns of targets, but also intensity losses play a role. This impact is direct as it affects the S/N (signal to noise ratio) of targets, whereas horseshoe bats are mainly worried about their F/C (flutter to clutter ratio), which is affected only indirectly by absorption losses. The relationship between signal design and quantitative performance in Vespertilionidae is still matter of investigation, however, signal design is adapted so strongly and instantaneously to the environment that nobody doubts a strong causal relationship. QCF frequency is part of this relationship. If a relationship is present, QCF frequency tells us something about the specialisation of the species. This is quite different from horseshoe bats where a CF call of 80 kHz tells us very little about the specialisation of a species compared to one calling at 100 kHz. This, in turn, may well be the reason why horseshoe bats have more freedom to “express” their identity as group, geographic location and so on in the frequency they use.