SEARCH

SEARCH BY CITATION

Keywords:

  • cochlear size;
  • inner ear;
  • basilar membrane;
  • scaling;
  • auditory epithelium

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

The primate cochlea is a membranous, fluid-filled receptor organ that is specialized for sound detection. Like other parts of the inner ear, the cochlea is contained within the bony labyrinth of the petrous temporal bone. The close anatomical relationship between the bony cochlear labyrinth and the membranous cochlea provides an opportunity to quantify cochlear size using osteological specimens. Although mechanisms of cochlear frequency analysis are well studied, relatively little is known about the functional consequences of interspecific variation in cochlear size. Previous comparative analyses have linked increases in basilar membrane length to decreases in both the high and low frequency limits of hearing in mammals. However, these analyses did not consider the potentially confounding effects of body mass or phylogeny. Here, we present measurements of cochlear labyrinth volume in 33 primate species based on high-resolution computed tomography. These data demonstrate that cochlear labyrinth volume is strongly negatively allometric with respect to body mass. Scaling of cochlear volume in primates is very similar to scaling of basilar membrane length among mammals generally. Furthermore, an analysis of 10 primate taxa with published audiograms reveals that cochlear labyrinth volume is significantly negatively correlated with the high frequency limit of hearing. This result is independent of body mass and phylogeny, suggesting that cochlear size is functionally related to the range of audible frequencies in primates. Although the nature of this functional relationship remains speculative, our findings suggest that some hearing parameters of extinct taxa may be estimated using fossil petrosals. Anat Rec, 292:765–776, 2009. © 2009 Wiley-Liss, Inc.

The mammalian inner ear is a complex series of fluid-filled receptor organs with varying sensory functions (Purves et al., 2008). Three receptor endorgans (the semicircular canals) detect angular accelerations of the head and help to stabilize the eyes and head via the vestibulo-ocular and vestibulo-collic reflexes. Two additional endorgans (the utricle and sacculus) primarily detect linear accelerations of the body and tilting of the head relative to the plane of gravity. The third component of the mammalian inner ear (the cochlea) detects sounds transmitted to the inner ear fluids by the external and middle ears. These six interconnected receptor organs together comprise the “membranous labyrinth” of the inner ear. The membranous labyrinth is suspended within the petrous temporal bone inside a network of fluid-filled bony voids, collectively called the “bony labyrinth”. The mammalian bony labyrinth includes five discrete spaces: one for each semicircular canal, one for the utricle and the sacculus, and one for the cochlea.

The close anatomical relationship between the bony labyrinth and the membranous labyrinth it contains has permitted paleontologists to reconstruct various dimensions of the inner ear in extinct taxa. In particular, the bony labyrinth has been used to estimate the size and orientation of the semicircular canals in comparative studies of locomotor behavior (Spoor et al., 1994, 1996, 2002, 2003, 2007; Witmer et al., 2003; Alonso et al., 2004; Rook et al., 2004; Clarke, 2005; Georgi, 2006; Boyer and Georgi, 2007; Silcox et al., in press). Because the utricle and sacculus are enclosed within a single bony cavity (the vestibule), it is not possible to estimate the dimensions of these endorgans with a similar degree of precision. However, the soft tissues of the mammalian cochlea are sequestered in a closely conforming, bony spiral canal (henceforth, the “cochlear labyrinth”) that is confluent with the vestibule.

The cochlear labyrinth is subdivided by soft tissues into three discrete spaces: the scala vestibuli, scala tympani, and scala media. The scala vestibuli and scala tympani are filled with the same fluid (perilymph) that bathes the external surface of the utricle and sacculus in the vestibule. By contrast, the scala media (or “cochlear duct”) is continuous with the other parts of the membranous labyrinth via a small communicating duct (ductus reuniens). Accordingly, the scala media contains the same fluid (endolymph) that fills the utricle, sacculus, and semicircular canals. The scala media is separated from the other cochlear scalae by two membranous partitions: the vestibular (=Reissner's) membrane (between the scala media and scala vestibuli) and the basilar membrane (between the scala media and scala tympani). The spiral organ of Corti, which is responsible for sound transduction in the cochlea, rests within the scala media and is anchored to the surface of the basilar membrane. When sounds are transmitted to the cochlear fluids by the auditory ossicles, transverse waves are produced on the basilar membrane that travel from the cochlear base (near the stapes) toward the cochlear apex. The location at which these traveling waves peak is dependent on the frequency of the stimulus, with lower frequency sounds generating traveling waves that propagate further along the basilar membrane than higher frequency sounds (Purves et al., 2008).

Although not all cochlear dimensions (e.g., the relative sizes of the three scalae) can be estimated using bony markers, the cochlear labyrinth can be used to reconstruct the total size of the cochlea and (with less certainty) the dimensions of the basilar membrane. As a result, the anatomy of the cochlear labyrinth has figured prominently in analyses of the phylogeny and hearing abilities of Mesozoic mammals (Graybeal et al., 1989; Luo and Ketten, 1991; Rosowski and Graybeal, 1991; Meng and Fox, 1995; Meng and Wyss, 1995; Fox and Meng, 1997; Hurum, 1998). Similarly, the cochlear labyrinths of fossil cetaceans have been used to discriminate between species specialized for high and low frequency hearing (Fleischer, 1976; Ketten, 1992; Luo and Eastman, 1995; Geisler and Luo, 1996; Luo and Marsh, 1996). More recently, the anatomy of the bony labyrinth has been used as an indicator of hearing abilities in australopithecines (Moggi-Cecchi and Collard, 2002), fossil archosaurs (Alonso et al., 2004; Gleich et al., 2005), and nonmammalian amniotes (Walsh et al., 2008).

Use of the cochlear labyrinth to make inferences about mammalian hearing abilities is based on the observation that the gross dimensions of the cochlea and its constituent tissues are correlated with the range of frequencies that can be detected by a species (West, 1985; Echteler et al., 1994). West (1985) demonstrated that the length of the basilar membrane is significantly negatively correlated with both the high and low frequency limits of hearing. In other words, as basilar membrane length increases in mammals, the range of audible frequencies shifts downward into relatively lower frequencies. As a result, mammals with absolutely long basilar membranes tend to have comparatively good low-frequency hearing, while mammals with absolutely short basilar membranes have comparatively good high-frequency hearing. Elephants, for example, have a basilar membrane length of about 60 mm and a range of audible frequencies between 0.18 and 10.5 kHz (West, 1985). By contrast, the house mouse has a basilar membrane length of only 7 mm, and a range of audible frequencies between 2.7 and 79 kHz (West, 1985).

Similar correlations between basilar membrane length and hearing thresholds were reported by Echteler et al. (1994) for mammals with “unspecialized” cochleas. However, these authors also demonstrated that species with substantially nonlinear cochlear frequency-place maps (“hearing specialists”) do not conform to this general mammalian trend (Echteler et al., 1994). From a practical standpoint, these results suggest that the dimensions of the cochlea may be used to estimate the high and low frequency limits of hearing for most mammals (West, 1985; Echteler et al., 1994). By contrast, the hearing abilities of taxa with acoustic foveae (e.g., dolphins and horseshoe bats) and/or substantial discontinuities in basilar membrane dimensions (e.g., mole rats) are more difficult to predict (Echteler et al., 1994). According to the criteria of Echteler et al. (1994), primates may be considered “hearing generalists”. None are known to possess acoustic foveae, none are specialized for echolocation or seismic hearing, and all species that have been studied exhibit typical cochlear frequency-place maps that lack plateaus or discontinuities (Greenwood, 1990). As a result, it is theoretically possible to derive predictions about the hearing abilities of primates based on the anatomy of the cochlea and cochlear labyrinth. Specifically, the findings of West (1985) and Echteler et al. (1994) suggest that increases in the length of the basilar membrane and resulting increases in the size of the cochlea should be correlated with a downward shift in the range of audible frequencies.

The goals of this analysis are twofold. First, we seek to document the allometric relationship between cochlear labyrinth volume and body mass in a broad comparative sample of 33 primate species. Empirical documentation of this scaling relationship is necessary to provide a “baseline” expectation for how cochlear size changes with increases or decreases in body mass. Second, we examine the relationship between cochlear labyrinth volume and hearing abilities using a subset of 10 primate taxa with published behavioral audiograms. In so doing, we seek to resolve the question of how changes in overall cochlear size may influence primate hearing. This analysis thus has important implications both for understanding functional anatomy of the primate cochlea and for reconstructing the sensory adaptations of extinct primate species. If, like basilar membrane length, cochlear volume is correlated with specific hearing abilities, then it may be possible to estimate some parameters of primate audiograms using suitably preserved fossil petrosals.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

CT Scanning

Measurements of cochlear labyrinth volume were collected for 33 primate species (10 lorisiforms, 12 lemuriforms, 11 haplorhines) using high-resolution X-ray computed tomography (μCT). μCT scans for all taxa except Eulemur fulvus, Callithrix jacchus, Saimiri sciureus, Chlorocebus aethiops, Pan troglodytes and Homo sapiens were produced at the Pennsylvania State University Center for Quantitative X-ray Imaging. These images were 16-bit unsigned data and were processed using a strip-2-raw program created by Nathan Jeffery (University of Liverpool). The six remaining μCT scans were produced at the University of Texas High Resolution X-ray CT Facility. These images were also 16-bit unsigned data but because no processing was performed, 8-bit conversion was required to import the data into imaging software. Most μCT scans were made as a series of slices through the cochlear labyrinth running approximately along the rostrocaudal axis of the petrosal. However, interspecific variation in the orientation of the modiolar axis, independent variation in the orientation and curvature of the basal turn of the cochlea, the difficulty of determining these parameters a priori by examining petrosal surface anatomy, and constraints imposed by scanner configuration and specimen dimensions precluded the possibility of scanning specimens in identical orientations. Resolution of μCT scans ranged between 18 and 47 μm in the x/y axes, and between 23 and 65 μm in the z axis.

Delimitation of the Cochlear Labyrinth

To distinguish the cochlear labyrinth from the vestibule, slices from each μCT image stack that did not include the cochlear labyrinth were first deleted. The membranous cochlear duct (= scala media) begins near the bony anchoring point of the basal-most basilar membrane (Echteler et al., 1994). This point, located between the oval and round windows, may be recognized in μCT images as the first appearance of a gap between the primary and secondary osseous spiral laminae (henceforth “basilar gap”). These spiral laminae anchor the inner (modiolar) and outer (abneural) edges of the base of the basilar membrane, respectively. Similarly, the base of the cochlear scala tympani is visible in μCT data as the bony limbus of the round window. In living specimens, the membrane closing the round window bulges into the middle ear space as the stapedial footplate presses into the oval window during sound transmission. By contrast, no bony landmarks are available to precisely delineate the boundary between the vestibule (containing the utricle and sacculus) and the cochlear scala vestibulae. The base of the cochlear labyrinth may therefore be precisely identified using bony markers associated with the scala media and scala tympani but not the scala vestibulae. Accordingly, the beginning of the cochlear labyrinth (and thus the first slice in each data file) was identified using the first appearance of either the basilar gap or the round window. Because of interspecific variation in cochlear anatomy and/or variation in the orientation of the cochlea during scanning, the first appearance of the round window was found to occur before, after, or in the same slice as the first appearance of the basilar gap. The end of the cochlear labyrinth was readily identified as the last slice in which the cochlear apex was visible.

Data Processing

Data were processed using ImageJ version 1.34k (Abramoff et al., 2004). Scans were cropped from the original size to a box that tightly enclosed the cochlear labyrinth. Three to four series of bounding lines were drawn with the line tool to close off openings in the cochlear labyrinth or to reinforce air-bone boundaries. All bounding lines were 1-pixel thick and white in color.

The first series of bounding lines were drawn between the two edges of the round window (Fig. 1a). These lines were drawn in every slice in which the round window was open, thus providing a clear separation between the cochlear labyrinth and the tympanic cavity. The second set of bounding lines separated the cochlear labyrinth from the vestibule. In the first slice of each scan in which the basilar gap was visible, a bounding line was drawn from the flexure at the base of the primary osseous spiral lamina to the nearest edge of the oval window (Fig. 1a). When this flexure was not distinct, the bounding line was drawn from the base of Rosenthal's canal (Fig. 1b). A similar line was drawn in successive slices until the confluence between the cochlear labyrinth and vestibule began to visibly narrow. At this point, the bounding line was moved from its initial orientation (primary lamina to oval window) to connect the narrowest point of the confluence between the cochlear labyrinth and vestibule (Fig. 1c). To assist with manual editing of data files (see below), additional bounding lines were drawn as needed along the inner surfaces of the internal acoustic meatus and/or modiolus (Fig. 1c,d). These bounding lines helped to close off the numerous small nerve foramina in the porous bone separating the cochlear labyrinth from the internal acoustic meatus and modiolus.

thumbnail image

Figure 1. Successive CT slices through the petrosal of an aye-aye (Daubentonia madagascariensis) illustrating the procedures for drawing bounding lines around the cochlear labyrinth. The bounding line separating the cochlear and vestibular portions of the bony labyrinth is marked with an asterisk (*). In slice A, this bounding line runs from the edge of the oval window (still occupied by the stapes) to the flexure at the base of the primary osseous spiral lamina. In slice B, this line runs from the oval window to Rosenthal's canal at the base of the primary osseous spiral lamina. Abbreviations: CL, cochlear labyrinth; G, basilar gap, between primary and secondary osseous spiral laminae; IAM, internal acoustic meatus; Mo, modiolus; Pr, canal for promontory artery; RW, bounding line closing round window; St, stapes or canal for stapedial artery; VL, vestibular labyrinth.

Download figure to PowerPoint

Calculation of Cochlear Labyrinth Volume

The boundary between the air-filled space of the cochlear labyrinth1 and the surrounding bone was estimated using the half-maximum height (HMH) method, in which the highest and lowest CT numbers on either side of an interface are averaged (Ullrich et al., 1980; Spoor, 1993; Spoor and Zonneveld, 1995). In μCT scans of the petrosal, however, the density of the bone surrounding the modiolus (including the primary osseous spiral lamina) is lower than that of the compact bone surrounding the rest of the cochlear labyrinth. As a result, it was not possible to pick a single HMH threshold that adequately identified the boundaries of the cochlear labyrinth throughout the entire scan (see also Coleman and Colbert, 2007). HMH thresholds chosen along the outer (abneural) edges of the cochlear labyrinth tended to overestimate cochlear labyrinth volume by including the osseous spiral laminae and portions of the bone surrounding the modiolus. Conversely, HMH thresholds chosen along the inner (modiolar) edges of the cochlear labyrinth often produced only partial selections of the cochlear labyrinth lumen and thus underestimated cochlear labyrinth volume. As a result, two separate HMH thresholds for each scan were calculated using the Plot Profile function in Image J. The first HMH threshold (“low threshold”) was calculated at the boundary between the cochlear labyrinth lumen and the primary osseous spiral lamina. This lower HMH value conservatively estimates cochlear labyrinth volume because it excludes the bone surrounding the modiolus as well as the osseous spiral laminae. A second HMH threshold was calculated at the outer (abneural) edge of the cochlear labyrinth where the lumen is adjacent to the very dense bone of the surrounding petrosal. However, this method produced very high HMH values and consequently included the spiral laminae and modiolus in selections of the cochlear labyrinth lumen. Accordingly, this second HMH value was averaged with the low threshold (calculated at the primary osseous spiral lamina) to produce a “high threshold” estimate of cochlear volume. The high threshold better identifies the outer (abneural) limits of the cochlear labyrinth than the low threshold, and liberally estimates cochlear labyrinth volume because it tends to include portions of the osseous spiral laminae and modiolar wall.

Cochlear labyrinth volume was calculated in Amira version 3.1.1 (Visage Imaging, Carlsbad, CA). Scans were imported into the Image Segmentation Editor window, and thresholds were applied to the entire scan. Air-filled spaces not included within the cochlear labyrinth (e.g., internal acoustic meatus, modiolus, and tympanic cavity) were manually deselected from each slice. Volumetric measurements were taken from the resulting selection of the cochlear labyrinth using the Tissue Statistics function. For each individual, a minimum cochlear labyrinth volume was calculated using the low threshold, and a maximum cochlear labyrinth volume was calculated using the high threshold (Table 1). Based on visual examination of 3D volumetric renderings, these values probably underestimate and overestimate (respectively) actual cochlear volume. Accordingly, the final estimate of cochlear labyrinth volume for each individual was obtained by averaging the maximum and minimum values (Table 1). All statistical tests were performed using this mean value for cochlear labyrinth volume.

Table 1. Cochlear labyrinth volume in primates
SpeciesCochlear labyrinth volume—lower estimate (mm3)Cochlear labyrinth volume—upper estimate (mm3)Cochlear labyrinth volume–mean (mm3)Body mass (kg)aHigh frequency limit (kHz)bLow frequency limit (kHz)bBest frequency (kHz)bHearing range (octaves)b
  • a

    Body masses were taken from Smith and Jungers (1997).

  • b

    Hearing parameters were taken from Heffner (2004).

Arctocebus calabarensis15.418.717.10.309    
Avahi laniger9.310.29.81.175    
Callicebus moloch16.220.118.20.988    
Callicebus torquatus17.721.319.51.245    
Callithrix jacchus8.910.49.70.320530 7 
Cebus apella23.028.325.73.085    
Cheirogaleus major4.66.05.30.4    
Cheirogaleus medius4.97.36.10.18    
Chlorocebus aethiops18.622.920.74.3450.0691.49.35
Daubentonia madagascariensis27.129.028.02.555    
Eulemur fulvus14.616.515.62.215430.07289.22
Euoticus elegantulus10.211.911.00.3    
Galago moholi6.89.38.10.18    
Galago senegalensis7.610.79.20.213650.092329.46
Galagoides alleni12.115.213.60.248    
Galagoides demidoff5.97.46.70.0615    
Hapalemur griseus11.813.112.40.709    
Homo sapiens76.483.980.259.4517.60.03149.15
Indri indri19.423.721.66.335    
Lemur catta9.513.611.52.21580.06789.76
Lepilemur sp.11.915.313.60.7655    
Loris tardigradus9.410.810.10.2665    
Macaca nigra24.229.326.77.6838.250.0282.9510.3
Microcebus murinus5.26.55.80.061    
Microcebus rufus3.84.64.20.0425    
Otolemur crassicaudatus12.718.415.61.15    
Otolemur garnetti13.417.315.40.764    
Pan troglodytes47.065.956.538.228.5 8 
Perodicticus potto11.614.913.30.833420.125168.39
Propithecus verreauxi11.716.013.93.545    
Saimiri sciureus11.214.512.80.7205430.188.75
Tarsius bancanus7.710.59.10.1225    
Tarsius syrichta9.211.410.30.1255    

Data Analysis

The data analysis began with an examination of the relationship between body mass and cochlear size. Two separate metrics of cochlear size were used: cochlear labyrinth volume (which is expected, but not known, to be correlated with hearing abilities) and basilar membrane length (which is known to be correlated with hearing abilities; West, 1985; Echteler et al., 1994). First, Pearson correlations were calculated to test the strength of the association between both inner ear variables and body mass. Second, ordinary least-squares (OLS) regressions were used to determine the percent of variation in cochlear labyrinth volume and basilar membrane length that could be explained as the result of variation due to body mass alone. Third, reduced major axis (RMA) regressions were calculated to examine the scaling relationship between both inner ear variables and body mass. Measurements of cochlear labyrinth volume in primates are listed in Table 1. Data on basilar membrane length for 38 mammalian species were taken from published literature and are provided in Table 2. Data on body mass follow Smith and Jungers (1997) for primates. Body mass for nonprimates follows Nowak (1991) or was taken from the original source that provided measurements of basilar membrane length (Table 2). When data on average body mass for both males and females were provided, the mean of the two values was used in making calculations. All variables were log10 transformed prior to analysis.

Table 2. Basilar membrane length in mammals
SpeciesCommon nameBody Mass (kg)Basilar membrane length (mm)Source
Arvicola terrestrisWater vole0.1310.5Lange et al., 2004
Balaena mysticetusBowhead whale100,00061.3Ketten, 1992
Bos taurusDomestic cattle50038Echteler et al., 1994
Cavia porcellusGuinea pig0.40620.5Begall and Burda, 2006
Chinchilla lanigerChinchilla0.4918.5Begall and Burda, 2006
Ctenomys talarumTuco-tuco0.1410.58Begall and Burda, 2006
Dipodomys merriamiMerriam's kangaroo rat0.059.83Echteler et al., 1994
Elephas maximusAsian elephant4,00060Echteler et al., 1994
Eubalaena glacialisNorthern right whale22,50055.6Ketten, 1992
Felis catusDomestic cat2.522.5Echteler et al., 1994
Fukomys anselliZambian mole rat0.0811.1Begall and Burda, 2006
Grampus griseusRisso's dolphin42540.5Ketten, 1992
Hipposideros fulvusFulvus roundleaf bat0.018.8Kössl and Vater, 1995
Hipposideros speorisSchneider's roundleaf bat0.019.2Kössl and Vater, 1995
Homo sapiensHuman7535Echteler et al., 1994
Lagenorhynchus albirostrisWhite-beaked dolphin10334.9Ketten, 1992
Macaca nemestrinaPigtailed macaque8.8525.6Greenwood, 1990
Megaderma lyraFalse vampire bat0.0489.9Kössl and Vater, 1995
Meriones unguiculatusMongolian gerbil0.0512.1Echteler et al., 1994
Microtus arvalisCommon vole0.0278.5Lange et al., 2004
Molossus aterBlack mastiff bat0.03714.6Kössl and Vater, 1995
Monodelphis domesticaShort-tailed opossum0.116.4Müller et al., 1993
Mus musculusHouse mouse0.016.8Echteler et al., 1994
Myotis lucifugusLittle brown bat0.0086.9Kössl and Vater, 1995
Oryctolagus cuniculusEuropean rabbit215.25Echteler et al., 1994
Pachyuromys duprasiFat-tailed gerbil0.0910.75Müller et al., 1991
Panthera oncaJaguar9033.3Burda et al., 1984
Panthera tigrisSumatran tiger106.335.5Burda et al., 1984
Phocoena phocoenaHarbor porpoise52.525.93Ketten, 1992
Pteronotus parnelliiParnell's mustached bat0.01214.3Kössl and Vater, 1995
Rattus norvegicusBrown rat0.32510.7Begall and Burda, 2006
Rattus rattusBlack rat0.212.1Begall and Burda, 2006
Rhinolophus ferrumequinumGreater horseshoe bat0.0216.1Echteler et al., 1994
Spalacopus cyanusCoruro0.0911.68Begall and Burda, 2006
Spalax ehrenbergiBlind mole rat0.14312.6Begall and Burda, 2006
Stenella attenuataSpotted dolphin112.536.9Ketten, 1992
Taphozous kachensisTomb bat0.0514.4Kössl and Vater, 1995
Tursiops truncatusBottlenose dolphin17540.65Ketten, 1992

To determine the relationship between cochlear size and hearing abilities in primates, four hearing parameters describing primate audiograms were taken from Heffner (2004). These parameters include the high frequency limit of hearing (highest audible frequency at 60 dB SPL), the low frequency limit of hearing (lowest audible frequency at 60 dB SPL), the best frequency of hearing (the frequency with the lowest absolute detection threshold), and the total hearing range (range of audible frequencies at 60 dB SPL, measured in octaves). Although the sample of cochlear labyrinth volumes includes data for 33 species of primates, audiograms have been published for only 9 of these species. These species include 2 lemuriforms (Eulemur fulvus, Lemur catta,), 2 lorisiforms (Galago senegalensis, Perodicticus potto), 2 platyrrhines (Callithrix jacchus, Saimiri sciureus), and 3 catarrhines (Chlorocebus aethiops, Pan troglodytes, Homo sapiens). The cochlear labyrinth sample also includes 1 macaque (Macaca nigra), and audiograms have been published for 4 macaque species (M. fascicularis, M. fuscata, M. mulatta, and M. nemestrina; Heffner, 2004). Accordingly, the hearing parameters of M. nigra were estimated using the mean for all 4 species of Macaca with known audiograms. Three-dimensional renderings of the 10 primate cochlear labyrinths used in functional analyses are shown in Fig. 2.

thumbnail image

Figure 2. This figure illustrates selected three-dimensional renderings of primate cochlear labyrinths. The 10 taxa shown here have associated audiograms, and were included in our functional analyses. All images have been scaled to the same approximate size, and are oriented so that the round window faces left.

Download figure to PowerPoint

The strength of the relationship between all four hearing parameters and cochlear labyrinth volume was first examined using Pearson correlations and OLS regressions. Because hearing abilities are known to be correlated with interaural distance (Heffner, 2004), Pearson correlations and OLS regressions were also calculated for all hearing parameters and body mass (Smith and Jungers, 1997). In the event that a hearing parameter was significantly correlated with both cochlear labyrinth volume and body mass, a partial correlation analysis was undertaken to examine the relationship between cochlear labyrinth volume and the hearing parameter while holding body mass constant. Such analyses involved calculating the residuals of both cochlear labyrinth volume and the hearing parameter from separate OLS regressions on body mass. The two sets of residuals were then compared using OLS regression, with residual cochlear labyrinth volume as the independent variable. In the event that the simple and partial correlation analyses for cochlear labyrinth volume a given hearing parameter were both significant, independent contrasts were used to examine the influence of phylogeny on the results. Independent contrasts were calculated using the PDAP module of Mesquite version 1.12 (Midford et al., 2003; Maddison and Maddison, 2006). The phylogenetic tree was derived from Xing et al. (2007), and branch lengths were set to N − 1 (N = number of terminal taxa in clade; Grafen, 1989). Pearson correlations were used to examine the relationship between independent contrasts for two variables.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

Scaling of the Primate Cochlea

Cochlear labyrinth volume is significantly positively correlated with body mass (r = 0.894; P < 0.001). This strong correlation is evident in Fig. 3, which presents a bivariate plot of log10 body mass (x-axis) by log10 cochlear volume (y-axis) for 33 primate species (Table 1). OLS regression reveals that body mass explains about 80% of the variation in cochlear volume. RMA regression demonstrates that cochlear volume is strongly negatively allometric with respect to body mass (expected slope for isometry = 1.0; observed slope = 0.36; confidence interval = 0.30–0.44). In other words, as body mass increases, the ratio of cochlear volume to body mass decreases. In this respect, scaling of cochlear volume in primates is very similar to scaling of basilar membrane length across mammals (Fig. 4). Like cochlear volume, basilar membrane length is significantly positively correlated with body mass (r = 0.939; P < 0.001). Similarly, regression analyses reveal that body mass explains a large proportion of the variation in basilar membrane length (∼88%) and that basilar membrane length is strongly negatively allometric with respect to body mass (Fig. 4; expected slope for isometry with basilar membrane length cubed = 1.0; observed RMA slope = 0.44; confidence interval = 0.39–0.50).

thumbnail image

Figure 3. Bivariate plot of log10 cochlear volume (in mm3) by log10 body mass (in kg) for 33 primate species. Data are taken from Table 1. For OLS regression: r2 = 0.80, F = 123.8, P < 0.001.

Download figure to PowerPoint

thumbnail image

Figure 4. Bivariate plot of log10 basilar membrane length (in mm; cubed) by log10 body mass (in kg) for 38 mammal species. Data are taken from Table 2. For OLS regression: r2 = 0.88, F = 267.2, P < 0.001.

Download figure to PowerPoint

Relationship Between Cochlear Labyrinth Volume and Hearing Abilities

Cochlear labyrinth volume is significantly negatively correlated with both the high and low frequency limits of hearing in primates (Table 3). By contrast, cochlear labyrinth volume is not significantly correlated with either best frequency or hearing range (Table 3). Body mass is also significantly negatively correlated with the high and low frequency limits of hearing, and is not correlated with best frequency or hearing range (Table 3). However, the strength of the correlation with high frequency limit is stronger for cochlear volume (r = −0.78; P = 0.0074) than for body mass (r = −0.68; P = 0.0321). Conversely, the strength of the correlation with low frequency limit is weaker for cochlear volume (r = −0.79; P = 0.0186) than for body mass (r = −0.84; P = 0.0089).

Table 3. Pearson correlations of cochlear size and body mass with hearing parameters
 High frequency limitLow frequency limitBest frequencyHearing range
Cochlear labyrinth volumer = −0.783r = −0.794r = −0.459r = 0.121
**P < 0.01*P < 0.05NSNS
n = 10n = 8n = 10n = 8
Body Massr = −0.675r = −0.841r = −0.552r = 0.273
*P < 0.05**P < 0.01NSNS
n = 10n = 8n = 10n = 8

OLS regression demonstrates that cochlear labyrinth volume can explain 61% of the variation in high frequency limit (Fig. 5a). However, if one taxon (Callithrix jacchus) that deviates substantially from the regression line is excluded, cochlear labyrinth volume can explain 88% of the variation in high frequency limit. By comparison, body mass can only explain 46% of the variation in high frequency limit if C. jacchus is included, and 74% if C. jacchus is excluded (Fig. 5b). Similarly, cochlear labyrinth volume and body mass can explain 63% and 71%, respectively, of the variation in low frequency limit (Fig. 6a,b).

thumbnail image

Figure 5. Bivariate plots showing correlates of high frequency limit for 10 primate species. a. Bivariate plot of log10 high frequency limit (in kHz) by log10 cochlear volume (in mm3). b. Bivariate plot of log10 high frequency limit (in kHz) by log10 body mass (in kg). Data are taken from Table 1.

Download figure to PowerPoint

thumbnail image

Figure 6. Bivariate plots showing correlates of low frequency limit for 8 primate species. a: Bivariate plot of log10 low frequency limit (in kHz) by log10 cochlear volume (in mm3). b: Bivariate plot of log10 low frequency limit (in kHz) by log10 body mass (in kg). Data are taken from Table 1.

Download figure to PowerPoint

The relationship between cochlear labyrinth volume and high frequency limit remains statistically significant when body mass is held constant in a partial correlation analysis (Fig. 7a). By contrast, cochlear volume is not significantly related to low frequency limit when body mass is held constant (Fig. 7b). When the relationship between cochlear labyrinth volume and high frequency limit is re-examined using independent contrasts, contrasts of the two variables remain significantly negatively correlated (r = −0.649; P < 0.05; N = 9). By comparison, body mass contrasts and high frequency limit contrasts are not significantly correlated (r = −0.494; NS; N = 9).

thumbnail image

Figure 7. Relationship between cochlear volume and hearing limits in primates, with body mass held constant. All variables are residuals from OLS regressions of log10 Y on log10 Body Mass. a: Bivariate plot of residual high frequency limit by residual cochlear volume. b: Bivariate plot of residual low frequency limit by residual cochlear volume. All calculations are based on data presented in Table 1.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

The results of this analysis provide further support for the conclusion that the dimensions of the cochlea influence hearing abilities in mammals. Previous research has shown that basilar membrane length is correlated with both the high and low frequency limits of hearing in mammals with unspecialized cochleas (West, 1985; Echteler et al., 1994). The present analysis demonstrates that cochlear size in primates (as estimated by the volume of the cochlear labyrinth) scales with body mass in a manner very similar to the scaling of basilar membrane length among mammals generally. In the comparative samples used here (Tables 1 and 2), cochlear labyrinth volume and basilar membrane length are each positively correlated with body mass (r = 0.894 and 0.939, respectively). Furthermore, both cochlear variables scale with strong negative allometry relative to body mass. Although cochlear labyrinth volume in primates demonstrates slightly greater negative allometry (RMA slope = 0.36) than basilar membrane length in mammals (RMA slope = 0.44) (Figs. 3 and 4), the RMA regression slope confidence intervals for both variables overlap. These similar scaling relationships support the expectation that cochlear size and basilar membrane length are closely linked. Indeed, it seems reasonable to expect that if selection acts to increase or decrease the length of the basilar membrane, then there should be correlated changes in cochlear volume. While caution in interpreting our data is warranted given the fact that the comparative samples for cochlear labyrinth volume (Table 1; primates) and basilar membrane length (Table 2; mammals generally) include different taxa, these results are consistent with the hypothesis that both cochlear variables should have a similar relationship with hearing abilities.

These expectations are largely borne out by our analysis of the relationship between cochlear labyrinth volume and hearing abilities in primate species with published audiograms (Heffner, 2004). Although cochlear labyrinth volume is not significantly correlated with either the best frequency of hearing (=frequency with lowest absolute detection threshold) or the total hearing range (= range of audible frequencies at 60 dB SPL), cochlear labyrinth volume is significantly negatively correlated with both the high and low frequency limits of hearing (Figs. 5a and 6a). In other words, as cochlear size increases, the range of audible frequencies shifts downward. These results are robust: even with the relatively small samples considered here, cochlear labyrinth volume alone can explain 61% of the variation in high frequency limit and 63% of the variation in low frequency limit. In this respect, our findings for cochlear size in primates closely match the results reported by West (1985) and Echteler et al. (1994) for basilar membrane length in a comparative sample of 9 mammalian species.2

One potential criticism of the analyses presented by West (1985) and Echteler et al. (1994) is that both failed to address the influence of body mass and phylogeny. Indeed, it has been known for decades that high frequency limit is correlated with head and body size (Masterton et al., 1969; Heffner and Heffner, 1992, Heffner, 2004). This correlation has typically been explained as a result of selection for efficient sound localization at different head sizes. According to Heffner:

Mammals with small heads (or, more precisely, short travel times for sound as it travels from one ear to the other) hear higher frequencies than mammals with large heads. The explanation for this relationship does not lie in the physical scaling of the auditory bulla and cochlea, with smaller middle and inner ears being associated with better high frequency hearing and larger ears being associated with better low-frequency hearing… In the case of high-frequency hearing, the explanation for the close correlation with head size… is that being able to detect high frequencies allows mammals to localize sound using pinna cues and spectral differences between the ears. (p. 1115; Heffner, 2004)

If this scenario is correct, then the size of the cochlea and length of the basilar membrane have no functional relationship with high frequency limit per se. In this case, the significant negative correlations between cochlear size variables (Table 1; West, 1985; Echteler et al., 1994) and high frequency limit would be the spurious byproduct of independent correlations between cochlear size, high frequency limit, and head/body size. The results of the present analysis, however, do not support this conclusion. While cochlear labyrinth volume and high frequency limit are both significantly correlated with body mass (Figs. 3 and 5b), the correlation between high frequency limit and cochlear volume (r = −0.78; P = 0.0074; Fig. 5a) is stronger than the correlation between high frequency limit and body mass (r = −0.68; P = 0.0321). Furthermore, the relationship between cochlear labyrinth volume and high frequency limit remains significant when independent contrasts are used to minimize the influence of phylogenetic effects. By comparison, independent contrasts of body mass and high frequency limit are not significantly correlated. More importantly, when a partial correlation analysis is used to hold the effects of body mass constant, cochlear labyrinth volume remains significantly correlated with high frequency limit (Fig. 7a). These results indicate not only that high frequency limit decreases with increasing absolute cochlear size, but that species with relatively large cochleas for their body size also have relatively low high frequency limits. In other words, at a given body size, species with smaller cochleas tend to have better high frequency hearing than species with larger cochleas.

The case for a functional relationship between cochlear size and low frequency limit is less convincing than that for high frequency limit. Although cochlear labyrinth volume is significantly negatively correlated with low frequency limit (r = −0.79; P = 0.0186), the correlation between body mass and low frequency limit is stronger (r = −0.84; P = 0.0089) (Fig. 6a,b). Furthermore, when a partial correlation analysis is used to hold body mass constant, cochlear labyrinth volume is no longer significantly correlated with low frequency limit (Fig. 7b).

These differences in the results for high frequency limit and low frequency limit beg the question of precisely how cochlear size might influence hearing abilities in mammals. Passive frequency analysis in the cochlea is generally attributed to the existence of a resonance gradient along the length of the basilar membrane, with the basal region of the basilar membrane having a higher resonant frequency than more apical regions (Echteler et al., 1994; Purves et al., 2008). This resonance gradient is thought to be largely dependent on variation in the mass and stiffness of the basilar membrane, leaving the mechanical significance of variation in basilar membrane length or the absolute size of the cochlea unclear. Although it is tempting to speculate that the mass of the cochlear fluids might also have a resonant effect on cochlear tuning, a great many factors influence cochlear mechanics (Dallos et al., 1996; Gummer, 2003) and a discussion of their potential relationship to cochlear size is well beyond the scope of the present analysis. Accordingly, while our data show a robust negative correlation between cochlear volume and high frequency limit in primates that is independent of body mass and phylogeny, the precise mechanism (or mechanisms) responsible for this relationship remains unknown.

It is also not immediately evident how interspecific variation in the high and low frequency limits of hearing influence primate ecology. Although the demands of sound localization are doubtless important (Masterton et al., 1969; Heffner and Heffner, 1992, Heffner, 2004), variation in habitat acoustics, diet, predation, and intraspecific communication may also exert a selective influence on hearing abilities (Morton, 1975; Waser and Brown, 1986; Zimmerman et al., 1995; de al Torre and Snowdon, 2002; Brumm and Slabbekoorn, 2005). Although these ecological relationships remain to be elucidated, the results of the present analysis suggest that the evolution of primate hearing abilities can be studied through an examination of the bony cochlear labyrinth. Cochlear labyrinth volume, in particular, can be used to estimate the high frequency limit of hearing. Similarly, Coleman and Boyer (2008) have recently reported that the length of the cochlea3 can be used to estimate low frequency sensitivity in euarchontans. These analyses hold out the possibility that some parameters of the audiogram can be reconstructed for fossil species with suitably preserved bony labyrinths. Ultimately, such conclusions may be linked to differences in auditory ecology, as with divergent cochlear specializations for echolocation in odontocetes and low frequency communication in mysticetes (Fleischer, 1976; Ketten, 1992; Luo and Eastman, 1995; Geisler and Luo, 1996; Luo and Marsh, 1996).

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

The authors acknowledge the assistance of Fred Spoor and Alan Walker, who generously provided access to μCT scans of primate petrosals. Likewise, Tim Ryan at the Penn State Center for Quantitative X-ray Imaging and Tim Rowe, Rich Ketcham, Matt Colbert, Jessie Maisano, and the staff of the University of Texas High Resolution X-ray CT Facility provided expert assistance with μCT scanning, data processing, and logistics. The initial concept for this paper grew from discussions of cochlear function with David Smith at Duke University in 1998. Without his influence, the project would not have been undertaken. This paper was also improved by discussions with Bill Henson, Marcelo Sánchez-Villagra, Liza Shapiro, and Mark Coleman. Two anonymous reviewers provided helpful comments for the revision of this manuscript. Adam Gordon kindly provided technical assistance regarding comparative phylogenetic methods.

  • 1

    Note that the cochlear labyrinth is air-filled in osteological specimens only.

  • 2

    Both West (1985) and Echteler et al. (1994) analyzed the same group of species with unspecialized cochleas (i.e., elephant, human, cow, chinchilla, Guinea pig, cat, rabbit, rat, and mouse). However, while West (1985) used the same 60 dB SPL threshold criterion to identify the high and low frequency limits of hearing as that employed here, Echteler et al. (1994) used a threshold criterion of only 30 dB SPL. As a result, the correlations between basilar membrane length and high and low frequency limits of hearing shown by Echteler et al. (1994) are slightly lower than those reported by West (1985).

  • 3

    Because this information is published in abstract form, it is not currently possible to evaluate how “cochlear length” (Coleman and Boyer, 2008) corresponds with the metrics of cochlear size evaluated here.

LITERATURE CITED

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED