2. Cetacean Ears
95
high-frequency hearing limit near 160kHz (Nachtigall et aI., Chapter 8).
Just as CFIFM bats have basal turn membrane anomalies and mole rats add
apical mass, all dolphins have anomalously narrow, thick membranes for
their length, and they add auxiliary stiffeners to the mix (Ketten 1984,1992).
With sufficient parameters, an accurate estimate can be calculated for frequency distributions for any animal. The first step is to determine the rules
for how ear structures scale from one animal to the next and how structural
parameters correlate with frequency. Multivariate analyses of the published
data on whale cochlear morphometrics data show frequency ranges and
peak spectra are reliably predicted (0.1 % confidence level) by a composite
of basilar membrane thickness/width ratios, laminae/length ratios, and turn
number (Table 2.1) (Ketten 1984). This composite boils down to a morphometric description of how stiffness varies with spiral position (Fig. 2.9).
Type I odontocetes have a basal ratio of greater than 0.8, outer laminar
support for greater than 60% of cochlear length, and peak frequency of
greater than 100 kHz. They also have low rise spirals of less than 2 turns.
Type II odontocetes have a basal ratio of less than 0.75, less than 30% outer
bony support, and a peak signal of less than 90 kHz. Type II cochleae are
steep spirals of greater than 2 turns. Type M spirals can be viewed two ways.
They are consistent proportionately with Type II formats but have lost highfrequency features. Alternatively they are simply very large generalists.
They do not, of course, have outer bony support elements or other stiffeners. Commensurate with their body type, mysticete basilar membranes are
exceptionally long. In terms of generalist fits, they are also exceptionally
broad and thin, implying very low stiffness and low to infrasonic hearing
abilities. At this point, primarily because of a lack of adequate cochlear duct
data, there are no data that show Type M ears to be anything except an
extended generalist.
These composite cochlear schematics, stripped to three parameters, are
the cetacean analog of Greenwood's human-derived formula for land
mammals. This accomplishes the first step in representing specialized ears:
establishing the minimal and/or optimal set of parameters needed for
comparing species.
It also provides the basis for the second step: formulating a media-blind
estimator of frequency ranges. Historically, researchers have progressively
added more parameters into the equation, but only rarely has there been
an attempt at retrospective analyses that selectively remove noncrucial elements (see Fay 1992). For mammalian ears, hearing range estimates for both
generalist and specialized ears are radically improved, up to a point, if more
than one parameter is used. That point is the watershed that differentiates
predominantly individual versus species-level adaptations (Ketten 1984). In
mammalian ears, based on comparisons of model versus audiometric data
for species with both available, two functionally related parameters, thickness and width, are sufficient (Ketten and Wartzok 1990). Further additions
will improve the tails of the hearing range estimates, but the ratio of thick-
95
high-frequency hearing limit near 160kHz (Nachtigall et aI., Chapter 8).
Just as CFIFM bats have basal turn membrane anomalies and mole rats add
apical mass, all dolphins have anomalously narrow, thick membranes for
their length, and they add auxiliary stiffeners to the mix (Ketten 1984,1992).
With sufficient parameters, an accurate estimate can be calculated for frequency distributions for any animal. The first step is to determine the rules
for how ear structures scale from one animal to the next and how structural
parameters correlate with frequency. Multivariate analyses of the published
data on whale cochlear morphometrics data show frequency ranges and
peak spectra are reliably predicted (0.1 % confidence level) by a composite
of basilar membrane thickness/width ratios, laminae/length ratios, and turn
number (Table 2.1) (Ketten 1984). This composite boils down to a morphometric description of how stiffness varies with spiral position (Fig. 2.9).
Type I odontocetes have a basal ratio of greater than 0.8, outer laminar
support for greater than 60% of cochlear length, and peak frequency of
greater than 100 kHz. They also have low rise spirals of less than 2 turns.
Type II odontocetes have a basal ratio of less than 0.75, less than 30% outer
bony support, and a peak signal of less than 90 kHz. Type II cochleae are
steep spirals of greater than 2 turns. Type M spirals can be viewed two ways.
They are consistent proportionately with Type II formats but have lost highfrequency features. Alternatively they are simply very large generalists.
They do not, of course, have outer bony support elements or other stiffeners. Commensurate with their body type, mysticete basilar membranes are
exceptionally long. In terms of generalist fits, they are also exceptionally
broad and thin, implying very low stiffness and low to infrasonic hearing
abilities. At this point, primarily because of a lack of adequate cochlear duct
data, there are no data that show Type M ears to be anything except an
extended generalist.
These composite cochlear schematics, stripped to three parameters, are
the cetacean analog of Greenwood's human-derived formula for land
mammals. This accomplishes the first step in representing specialized ears:
establishing the minimal and/or optimal set of parameters needed for
comparing species.
It also provides the basis for the second step: formulating a media-blind
estimator of frequency ranges. Historically, researchers have progressively
added more parameters into the equation, but only rarely has there been
an attempt at retrospective analyses that selectively remove noncrucial elements (see Fay 1992). For mammalian ears, hearing range estimates for both
generalist and specialized ears are radically improved, up to a point, if more
than one parameter is used. That point is the watershed that differentiates
predominantly individual versus species-level adaptations (Ketten 1984). In
mammalian ears, based on comparisons of model versus audiometric data
for species with both available, two functionally related parameters, thickness and width, are sufficient (Ketten and Wartzok 1990). Further additions
will improve the tails of the hearing range estimates, but the ratio of thick-