THE ANATOMICAL RECORD PART A 281A:1123–1137 (2004) Primate Auditory Diversity and Its Influence on Hearing Performance 1 MARK N. COLEMAN1* AND CALLUM F. ROSS2 Interdepartmental Doctoral Program in Anthropological Sciences, Stony Brook University, Stony Brook, New York 2 Department of Organismal Biology and Anatomy, University of Chicago, Chicago, Illinois ABSTRACT The auditory region contains numerous structures that have proven useful for phylogenetic classification at various taxonomic levels. However, little work has been done in primates relating differences in morphology to variations in hearing performance. This study documents anatomical and physiological distinctions within primates and begins to address the functional and evolutionary consequences of these and other auditory features. The dimensions of the outer ear (pinna) were measured in cadaveric specimens representing nearly every primate family and used to calculate a shape ratio (height/width). It was found that nonanthropoids have a significantly higher ratio than anthropoids, although the actual height was not found to differ. This indicates that most nonanthropoids have ears that are tall and narrow, whereas monkeys and apes are characterized by ears with more equal height and width dimensions. Eardrum area, stapedial footplate area, and ossicular lever arm lengths were measured in dried specimens to calculate an impedance transformer ratio. A distinction was found between anthropoids and strepsirrhines, with the latter group having a transformer ratio indicative of a higher percentage of acoustic energy transmission through the middle ear. Audiogram data were gathered from the literature to analyze hearing sensitivity and it was found that platyrrhines illustrate more low-frequency sensitivity than like-sized lorisoids. The effects of intraspecific variation on the audiogram results were also examined and were found to produce similar results as the analysis using species mean threshold values. Lastly, correlations between morphological and audiogram variables were examined. Several measures of hearing sensitivity were found to be correlated with pinna shape but correlations with middle ear transmission properties were weaker. In addition to using traditional statistical techniques, phylogenetic corrective methods were applied to address the problem of statistical nonindependence of the data and the results of both analyses are compared. These findings are discussed with respect to how sensory adaptations and phylogenetic history may be related to the current radiation of primates. © 2004 Wiley-Liss, Inc. Key words: primate hearing; outer ears; middle ears; audiogram; phylogeny Primates express numerous modifications on the fundamental mammalian middle ear morphology and these differences have proven useful for phylogenetic classification at various taxonomic levels (MacPhee and Cartmill, 1986). Among extant strepsirrhines, most Malagasy primates are characterized by a free-floating tympanic ring surrounded by a single relatively large tympanic cavity while lorisoids have a tympanic ring that is fused to the lateral wall of a smaller tympanic cavity. Lorises appear to maintain a substantial effective cavity volume by having additional pneumatic spaces off of the epitympanic recess. Extant haplorhines are similar to lorises in possessing epitympanic sinuses (except tarsiers) (MacPhee and Cartmill, 1986) and a fused tympanic ring but are further © 2004 WILEY-LISS, INC. distinguished by the presence of a diverticulum off the eustachian tube called the anterior accessory cavity. In Presented at the Symposium on Primate Sensory Systems at the 2004 American Association of Physical Anthropologists (AAPA) Meeting in Tampa, Florida. *Correspondence to: Mark N. Coleman, Academic Tower A, T-8069, Stony Brook University, Stony Brook, NY 11794. Fax: 631-444-3947. E-mail: mcoleman@ic.sunysb.edu Received 20 May 2004; Accepted 1 July 2004 DOI 10.1002/ar.a.20118 Published online 6 October 2004 in Wiley InterScience (www.interscience.wiley.com). 1124 COLEMAN AND ROSS Despite the well-documented abundance of diversity in the outer and middle ears of primates, the functional implications of these differences are not well understood (Cartmill, 1975; Fleagle, 1999). A better understanding of how diversity in ear structure affects hearing performance will shed considerable light on the sensory adaptations that led to the radiation of extant primates and their fossil relatives. The goals of this study are to test for taxonomic differences in primate middle and outer ear morphology and estimate their functional characteristics using models based on acoustical theory to investigate how well these model predictions relate to actual hearing performance. Similar lines of investigation have been applied to other vertebrate groups (e.g., rodents and felids), but primates have received only limited attention in this regard, with the few relevant studies narrowly focused on only a few taxa (Packer and Sarmiento, 1984; Masali et al., 1992; Hemilä et al., 1995; Moggi-Cecchi and Collard, 2002). Theoretical Background Fig. 1. Basic pinna morphology scaled to approximately the same size for various primates representing each family/subfamily. Note the elongated auricular lamina characteristic of most prosimians (first two rows minus Callithrix) that results in a relatively tall and narrow pinna compared with most anthropoids (last two rows plus Callithrix). tarsiers, this cavity is nontrabeculated, while in all other haplorhines, it is filled with trabeculae. The outer ears of primates also show considerable diversity in morphology and mobility that often follows phylogenetic patterns. Ceboids, lemuroids, and most lorisoids have an ear canal that is almost entirely cartilaginous while in cercopithecoids, hominoids, and tarsioids the canal is composed mostly of bone formed by a tubular expansion of the ectotympanic. Primate pinnae are highly variable in overall shape as illustrated by Figure 1, but can generally be characterized by one of two major patterns. Most prosimians have a superiorly elongated auricular lamina (helix and antihelix) that is commonly ovate or conical in shape while anthropoids typically show a more subquadrate outline with an involuted helix (Hershkovitz, 1977). This results in prosimians having what appear to be tall and narrow pinnae compared with those of monkeys and apes that are more equal in width and height dimensions. In addition, the majority of prosimians have larger pinnae than those of all anthropoids, with aye-ayes having the largest ears and orangutans having the smallest ears relative to head size (Schultz, 1973). Prosimians also show better developed auricular musculature (Huber, 1930; Lightoller, 1934; Schultz, 1969; Burrows and Smith, 2003) and generally more mobile pinnae than anthropoids (Hill, 1955; Hershkovitz, 1977). Several models have been developed to describe the functional characteristics of the outer ear, but one simple approach relates the shape of the outer ear to its frequency selectivity and directionality. Outer ears with long and narrow dimensions should show a decrease in lowfrequency reception while ears that are wider or larger will be sensitive to low- as well as high-frequency sounds and will provide more directional cues at low frequencies (Rosowski, 1994). Although intuitively it might seem advantageous for all animals to maximize sensitivity to as wide a range of frequencies as possible, specialization might entail trade-offs. For example, a species that relies on high-frequency reception for survival might find it advantageous to avoid reception of low-frequency sounds that could mask more important high-frequency sounds. The main role of the middle ear is to help overcome the impedance mismatch that results from the higher acoustic impedance of perilymph in the inner ear compared to that of air (⬃ 134:1) (Zwislocki, 1965). The two primary mechanisms that have been proposed to assist in this task are the ossicular lever arm ratio (the lever action that results from the uneven lengths of the manubrium of the malleus and the long process of the incus) and the areal convergence ratio (the increase in pressure that results from the larger surface area of the tympanic membrane relative to the stapedial footplate; Fig. 2). Using these two ratios, it is possible to calculate the impedance transformer ratio (ITR) of the middle ear, given by the formula ITR ⫽ (As/(2/3 Ad))*(Li/Lm)2, where As is the surface area of the stapedial footplate, Ad is the area of the tympanic membrane, Li is the lever arm of the incus, and Lm is the lever arm of the malleus. The ITR is often considered an ideal transformer ratio because it ignores the intrinsic impedance of the components of the auditory system itself (Dallos, 1973), and several researchers have found a lack of association between the ITR or its component ratios (e.g., areal convergence ratio) and measures of auditory sensitivity (Lay, 1972; Rosowski, 1994). However, it provides a useful starting point for analyzing middle ear function and continues to be used by investigators as a proxy for estimating middle AUDITORY DIVERSITY IN PRIMATES 1125 MATERIALS AND METHODS Fig. 2. Schematic representation of the areal convergence and ossicular lever arm ratios used to calculate the impedance transformer ratio. Ad represents the area of the tympanic membrane, As the surface area of the stapedial footplate, Lm the lever arm of the malleus, and Li the lever arm of the incus. Methods used for obtaining the measurements are described in text. ear performance (Webster and Webster, 1975; Hunt and Korth, 1980; Masali et al., 1992). Using the ITR, it is possible to estimate the theoretical maximum percentage of acoustic transmission (T) at peak performance through the middle ear using the formula To investigate the functional consequences of differences in primate ear morphology, outer ear shape, middle ear impedance matching performance, and audiograms were assayed in 49 genera of primates. These taxa fall into two monophyletic suborders, Strepsirrhini (lemurs, lorises, and galagos) and Haplorhini (tarsiers and anthropoids). The terms “Prosimii” and “prosimians” are used to refer to a paraphyletic grouping of strepsirrhines and tarsiers, i.e., nonanthropoid primates. Complete data sets could not be gathered for all genera; audiogram data were only available for 10 genera, outer ear data were available for 36 genera, and middle ear data were available for 34 genera. Morphometric data were gathered from specimens in the collections at the American Museum of Natural History, Field Museum of Natural History, and National Museum of Natural History. Whenever a structure from both ears of a single specimen was measurable, the average value of both ears was used. Morphometric data were pooled at the generic level due to the difficulty in obtaining adequate sample sizes for various species. In addition to the morphometric measures on auditory structures described below, 17 cranial measurements (Table 1) were taken on the dried skulls in order to calculate a geometric mean of skull size for use in size evaluation. Geometric mean (GM) equals the Nth root of the product of N variables ((GM ⫽ N 冑 [CM1 ⫻ CM2 ⫻ . . . ⫻ CMN]). The general combination of cranial measurements used to calculate GM has been found to produce consistent analytical results while maximizing statistical power (Coleman, 2003). Outer Ear T ⫽ 4r / (r ⫹ 1) 2 where r ⫽ Z1 / Z2, the ratio of the specific acoustical impedance at the eardrum (Z1) to the characteristic specific acoustical impedance of air (Z2 ⫽ 41.5 dynes sec/cm3). Z1 is determined by multiplying the specific acoustical impedance of the inner ear by the ITR. The actual hearing performance that results from the combined effects of the outer, middle, and inner ears can be evaluated through behavioral testing in several ways, including determining absolute auditory thresholds, localization acuity, and amplitude, temporal, or frequency difference limens (smallest detectable differences). Absolute auditory thresholds were examined in this study since they are considered the most fundamental evaluation of audition (Stebbins, 1975) and provide basic measures of hearing such as high- and low-frequency sensitivity, frequency of greatest sensitivity, and overall range of audible frequencies. Threshold values are graphically represented as bivariate plots called audiograms with frequency measured in hertz (Hz) displayed on the abscissa and amplitude measured in decibels (dB) displayed on the ordinate. Audiograms are available in the literature for 17 species of primates, including four strepsirrhines, three platyrrhines, eight cercopithecines, and two hominoids (Fig. 3). To calculate a shape index (Psi) for the outer ear, measurements of the maximum height and width of the pinnae were taken to the nearest millimeter using the traditional landmarks shown in Figure 4: Psi ⫽ pinna height / pinna width. Since many primates do not have a true ear lobe, subaurale was determined as the most inferior aspect of the outer ear as it extends laterally from the side of the head. Due to the concern that drying and storage of museum skins may distort the size and shape of the outer ear, pinna dimensions were taken only on cadaveric specimens (204 ⫹ 2 anesthetized animals). The cadavers are housed at the National Museum of Natural History and were fixed in 10% buffered formalin and are stored in 70% ethanol. Middle Ear To calculate the ITR, estimates were obtained of the surface areas of the tympanic membrane and stapedial footplate and the lengths of the malleolar and incudal lever arms. The tympanic membrane is often deteriorated in most dried specimens, so the area was estimated by making a mold that preserves the impression of the tympanic ring (or the rarely preserved membrane) and then taking measurements on the impression. The mold was made by injecting polyvinylsiloxane into the lateral aspect of the middle ear cavity. The mold was then removed and sectioned under a microscope along the line of attachment of the tympanic membrane. The modified molds were then digitally photographed at a distance at least 12 times the maximum dimension to be measured to minimize the ef- 1126 COLEMAN AND ROSS Fig. 3. Audiograms for 17 species of primates representing every primate superfamily except Tarsioidea. Data taken from the original literature: Aotus trivirgatus (Beecher, 1974a); Callithrix jacchus (Seiden, 1957); Cercopithecus mitis (Brown and Waser, 1984); Cercopithecus neglectus and Chlorocebus aethiops (Owren et al., 1988); Galago senegalensis (Heffner et al., 1969); Lemur catta (Gillette et al., 1973); Macaca fuscata and Homo sapiens (Jackson et al., 1999); Macaca mulatta (Pfingst et al., 1978); Macaca nemestrina and Macaca fascicularis (Stebbins et al., 1966); Nycticebus coucang and Perodicticus potto (Heffner and Masterson, 1970); Pan troglodytes (Kojima, 1990); Papio cynocephalus (Hienz et al., 1982); Saimiri sciureus (Beecher, 1974b). 1127 AUDITORY DIVERSITY IN PRIMATES TABLE 1. Measurements taken on dried skulls used to calculate the geometric mean of skull size Measurement Skull length Skull width Facial width Orbital height Orbital depth Basicranial length Interaural distance Neurocranial length Neurocranial width Neurocranial height Palate width Palate length Ramus height Symphyseal height Symphyseal width Corpus height Corpus width Osteometric landmarks Region/apparatus Prosthion-inion Bi-zygion Bi-ectoconchion Orbitale inferiorus-orbitale superiorus Orbitale inferiorus-anterior optic canal Prosthion-basion Bi-ectotympanic Nasion-inion Bi-euryon Basion-vertex Bi-ecotomolare Prosthion-staphylion Gonion-condyle of ascending ramus Infradentale-gnathion Pogonion-posterior symphyseal border Superior-inferior border of corpus at M2 Mandibular corpus width at M2 Full skull Full skull Facial/visual Facial/visual Facial/visual Basicranial Basicranial/auditory Neural Neural Neural Palate/masticatory Palate/masticatory Mandible/masticatory Mandible/masticatory Mandible/masticatory Mandible/masticatory Mandible/masticatory Fig. 4. Measurements taken on cadaveric pinnae to calculate a height/width shape index. Drawing of Papio hamadryas outer ear. fects of parallax (Spencer and Spencer, 1985). Care was taken to place the plane of the tympanic ring impression parallel to the lens of the camera. Digital images were imported into Sigma Scan Pro 5.0 image measurement software, calibrated, and the surface area was calculated by tracing the perimeter (excluding the pars flaccida portion). The procedure is preferred over traditional methods that calculate area based on two perpendicular axes because it permits deviations from strictly round outlines to be incorporated into the estimate, allows the measurement of tympanic rings that are not fully visible externally, and creates a permanent mold that can be repeatedly measured. However, because this technique cannot be applied to taxa with a substantial bony ear canal, tympanic membrane estimates were not obtained for catarrhine or tarsier specimens. The surface area of the stapedial footplate was measured by taking a digital image of the footplate and measuring the image using the same protocols outlined for tympanic membrane molds. To increase the number of specimens for which areal convergence ratios could be calculated, the oval window was measured as a proxy for the stapedial footplate whenever possible. Preliminary investigation on specimens for which both measurements could be obtained shows a tight correlation and isometric relationship between these two measures (r2 ⫽ 0.961; slope ⫽ 0.996 ⫾ 0.079). Replication experiments were carried out to evaluate the precision of these digital measurement techniques and incorporated the potential variation associated with lens-to-object angle, parallax, calibration, and measurement error. Nonsignificant differences were found between two different measurement experiments of the same specimens for both tympanic membrane area (paired t-test, n ⫽ 54; P ⫽ 0.623) and stapedial footplate area (paired t-test, n ⫽ 18; P ⫽ 0.538). The lengths of the malleolar and incudal lever arms were measured using similar digital imaging and measurement techniques as described above. The ossicles were positioned so that the axis of rotation and the lever arms were parallel to surface of the lens. The lengths were determined by first drawing a line representing the axis of rotation from the short process of the incus through the long process of the malleus and then drawing perpendicular lines from this axis to the tips of the manubrium and long process of the incus. Although the malleus-incus complex rarely remains intact in the loose ossicles found in museum collections, the complex saddle-shaped articular surfaces of these bones should permit a good reconstruction of the orientations of the malleus and incus in their natural positions relative to each other. This supposition was verified in the following manner. First, articulated malleus-incus pairs from specimens from the Stony Brook Comparative Anatomical Museum were measured. Next, the pairs were separated and then rearticulated, with the aid of a microscope, on a piece of clay, then rearticulated 1128 COLEMAN AND ROSS ossicles were measured again. It was found that the articulated and rearticulated measurements did not significantly differ (paired t-test, n ⫽ 24; P ⫽ 0.885). To calculate the theoretical percentage of acoustic transmission through the middle ear (T), one needs an estimate of the specific acoustic impedance of the cochlea in addition to the ITR. The specific acoustic impedance of the cochlea is primarily determined by “the viscous drag of the cochlear fluids against the walls of the scalae” (Webster and Webster, 1975) and different animals will show some variation in cochlear impedance due to differences in the dimensions of the scalae. However, the specific acoustic impedance of the cochlea has not been determined for any nonhuman primates and only a few other mammalian species. Previous studies on great apes (Masali et al., 1992), dermopterans (Hunt and Korth, 1980), and heteromyid rodents (Webster and Webster, 1975) have used a value of 5,600 dynes sec/cm3 as an approximation for mammalian cochleae in general based on an estimate of cochlear impedance derived for humans (Zwislocki, 1965). However, subsequent measurements of the specific acoustic impedance of the inner ear suggest a value of 11,200 dynes sec/cm3 for humans (Zwislocki 1975), twice the previous estimate. In this study, a value of 5,600 dynes sec/ cm3 will be used to calculate the percentage of transmission (T1) in order to compare it to the results from previous studies and using 11,200 dynes sec/cm3 to derive a more accurate estimate of the theoretical percentage of acoustic transmission through the middle ear (T2). Fig. 5. Audiometric variables measured on each audiogram. These include frequency and threshold of the primary peak (lowest point on the audiogram), low-frequency cutoff (lowest audible frequency at 40 dB), high-frequency cutoff (highest audible frequency at 60 dB), total audible range (measured in octaves at 40 dB), total area of the audible field (area encompassed within threshold values below 60 dB), low area (area below 1,000 Hz), middle area (area between 1,000 and 8,000 Hz), and high area (area above 8,000 Hz). Additional measurement information available in text. Audiograms Audiograms were analyzed by measuring nine audiometric variables similar to those used in previous studies (Masterson et al., 1969; Rosowski and Graybeal, 1991): frequency and threshold of the primary peak, low-frequency cutoff at 40 dB, high-frequency cutoff at 60 dB, total range in octaves at 40 dB, total area of the audible field below 60 dB, low area, middle area, and high area (Fig. 5). The cutoff points were established based on data available for all audiograms. The divisions between low, middle, and high areas were arbitrarily set at 1 and 8 kHz. Areal measurements were made by importing the raw audiogram data into IGOR PRO 4.04 wave measurement software. Analyses Tests for taxonomic effects. Initial examination of the data revealed obvious differences between haplorhines or anthropoids and strepsirrhines in many of the variables. The significance of these suborder differences was first tested using nonparametric Mann-Whitney U-tests, considered significant at P ⱕ 0.05. However, traditional statistical methods for the analysis of comparative data fail to take into account the nonindependence of species/ genus means due to phylogenetic relatedness (Felsenstein, 1985), resulting in inflated type I error rates and lowered statistical power (Garland et al., 1993). Methods are now available for the incorporation of patterns of phylogenetic relationship into statistical calculations of ANOVA, ANCOVA, correlation, and regression. These techniques are applied here using PDAP (Phenotypic Diversity Analysis Programs, version 6, 2002) (Garland et al., 1993) and the results are compared with those obtained using traditional parametric and nonparametric methods. The phylogenetic relationships among the primates examined in this study are illustrated in Figure 6. The topology and branch lengths in this tree differ slightly from those in Ross et al. (2004) as explained in Figure 6. The lack of complete data for all taxa meant that different subsets of the tree were used in different analyses. These trees were created by pruning from the tree those taxa for which the relevant data were not available. For each tree, the statistical adequacy of the branch lengths was tested using the diagnostics in PDAP. No branch length transformations were necessary. Tests for suborder effects on morphometric and audiometric data were performed using PDANOVA. The test statistic (F ratio) was calculated using PDSINGLE (although a standard statistical package could have been used) and the significance of this statistic was evaluated using Monte Carlo simulations of the traits on the phylogenetic tree using PDSIMUL. Ten thousand simulations were run under a speciational Brownian motion model. This model assumes all evolution occurs at speciation events, not along branches, so all branch lengths are set to unity. We used this simulation model because the lengths of the basal branches in Figure 6 are controversial at present. For each variable, boundary conditions were set to be slightly broader than the range seen in extant primates and the initial or starting value was the assumed primitive condition. These simulations produce null distributions of F ratios that were imported into MS Excel where the critical value of the F ratio at the 95th percentile was identified (hereafter referred to as Fphylo). Audiometric analyses. Several problems confound a phylogenetic analysis of the audiometric data: there are only 4 strepsirrhines with audiograms compared with 14 1129 AUDITORY DIVERSITY IN PRIMATES similarly sized species for which threshold data on several individuals are available (⬎ 3). Correlation analyses. The final step was to test the association between morphometric data and audiometric data to see if differences in auditory structures are correlated with differences in hearing performance. For this stage of the analysis, 10 audiograms were analyzed (the 10 for which the species also have complete information on outer ear shape and/or middle ear impedance transformer ratios). Since this abbreviated data set has a maximum n of nine, all correlations with a P value of ⱕ 0.10 were considered significant because such small samples sizes permit a more relaxed significance criterion (Olson and Miller, 1958). These tests were conducted using both traditional Spearman’s rank order correlations and phylogenetically independent contrasts in PDTREE with branch lengths transformed to unity. The degree to which the correlations between form and function agree with predictions based on auditory theory is scrutinized below. RESULTS Outer Ear Fig. 6. Phylogenetic tree of primates examined in this study. The PDTREE files are available from the authors on request. Numbers are ages of nodes in million years. Tree structure follows that used by Ross et al. (2004), with the following changes in node dates: Microcebus added as sister taxon to Mirza, diverging at 20 Myr (Yoder and Yang, 2004); age of Primates increased to 85 Myr; age of Anthropoidea increased to 50 Myr; age of Strepsirrhini increased to 68.5 Myr; age of Malagasy strepsirrhine ancestral node increased to 62 Myr; age of basal Cheirogaleidae node increased to 29 Myr; age of internal lemuriform node decreased to 42 Myr; age of Lemuridae decreased to 32; and age of basal Lorisiformes decreased to 40 Myr, all after Yoder and Yang (2004). The divergence of the Eulemur clade from the Hapalemur-Lemur clade was estimated at 28 Myr and of Hapalemur and Lemur at 17 Myr based on Yoder and Yang (2004). Yoder and Yang (2004) do not support the clade Indriidae, consisting of (Indri, Propithecus, Avahi), with Lepilemur as its sister taxon, so original clade dates from Ross et al. (2004) are used. The age of the Haplorhini node was increased to 67.5 Myr, midway between Anthropoidea (50) and Primates (85) nodes dated by Yoder and Yang (2004). Various anthropoid taxa were also trimmed from the tree and Nasalis was added to the Simias branch, diverging at 4.5 Myr, halfway from the stem node to the present. anthropoids; there is limited overlap in the body sizes of these groups; and several of the audiometric variables are correlated with body mass [body mass data from Smith and Jungers (1997)]. To control for these problems, a narrow allometric approach was used, considering only the taxa that are in the same size range. This limited the comparison to the three lorisoid and three platyrrhine audiograms. The audiometric data used in this study (and most other studies) represent mean threshold values calculated within species. These average threshold values may mask underlying intraspecific variation. Comparative analyses of mean audiograms would ideally include estimates of intraspecific variability; however, such data are seldom available. Here we present a comparison of the audiograms of Galago senegalensis and Callithrix jacchus, two Table 2 presents the generic averages for the variables used in the outer ear analysis. The pinna shape index (Psi) is plotted by family/subfamily in Figure 7A. The values for Cebus and Saimiri are plotted separately. This plot underscores the pattern observed in Figure 1 that the major distinction in primate outer ear shape is between anthropoids and all other primates. There is a significant difference in outer ear shape between these two groups (U ⫽ 9; P ⬍ 0.001) with anthropoids having lower index values indicative of a pinna shape that is more symmetrical compared with the relatively tall and narrow pinnae characterizing nonanthropoids. This difference in pinna shape index may be influenced more by differences in pinna width than pinna height. Comparisons of small-bodied primates suggest that nonanthropoids have narrower pinnae than anthropoids, although this difference is not quite statistically significant at P ⱕ 0.05 (F ⫽ 3.88; P ⫽ 0.06). Certainly, in this body size range, the values for pinna height show almost complete overlap. The suborder-level effects on Psi remain significant using phylogenetically adjusted critical values of the F ratio (F ⫽ 72.18; Fphylo ⫽ 44.05). Middle Ear Table 3 presents the generic averages for the variables used in the middle ear analysis. The percentage of acoustic transmission (T1), calculated using the traditional cochlear impedance value of 5,600 dynes sec/cm3, is presented in Figure 8 grouped by family/subfamily. Once again, subordinal level differences are apparent. The majority of anthropoids have transmission values between 81% and 92% while strepsirrhines show higher values ranging from 92% to nearly 100%. The exception to this dichotomy is the Callitrichinae, with all five genera having T1 values between 93% and 97%. Using the higher estimate of cochlear impedance (11,200 dynes sec/cm3), anthropoids exhibit T2 values ranging from 54% to 69%, strepsirrhines from 68% to 92%, and callitrichines with values between 70% and 76%. Despite the intermediate values for callitrichines, the difference between anthropoids and strepsirrhines remains significant when evaluated using standard nonparametric statistics, regardless 1130 COLEMAN AND ROSS TABLE 2. Height and width measurements in millimeters taken on primate outer ears used to calculate a pinna shape index (Psi)* Genus Alouatta Aotus Arctocebus Ateles Callimico Callithrix Cebus Cercocebus Cercopithecus Cheirogaleus Daubentonia Erythrocebus Euoticus Galago Galagoides Hylobates Lagothrix Leontopithecus Loris Macaca Microcebus Miopithecus Nasalis Nycticebus Otolemur Papio Perodicticus Presbytis Pygathrix Saguinus Saimiri Semnopithecus Tarsius Theropithecus Trachypithecus Varecia Total Pinna height (mm) Pinna width (mm) Pinna ratio (Psi) 33.9 ⫾ 2.33 (10) 27.6 ⫾ 1.58 (10) 25.0 (1) 33.0 ⫾ 3.62 (10) 25.5 ⫾ 0.71 (2) 24.4 ⫾ 1.41 (8) 35.0 ⫾ 2.29 (23) 33.5 ⫾ 2.12 (2) 33.1 ⫾ 3.80 (7) 17.3 ⫾ 0.29 (4) 75.1 (1) 43.7 ⫾ 4.89 (7) 25.3 ⫾ 0.28 (2) 38.8 ⫾ 2.80 (6) 24.1 ⫾ 1.34 (9) 31.2 ⫾ 2.95 (5) 29.8 ⫾ 1.89 (4) 27.0 ⫾ 0.63 (6) 23.8 ⫾ 1.50 (4) 38.1 ⫾ 3.59 (9) 22.5 ⫾ 0.71 (2) 28.8 ⫾ 1.72 (6) 34.5 ⫾ 3.00 (4) 23.0 ⫾ 2.18 (3) 46.3 ⫾ 11.2 (3) 50.1 ⫾ 4.94 (9) 24.0 (1) 32.5 ⫾ 2.12 (2) 37.4 ⫾ 2.61 (5) 19.0 ⫾ 0.82 (10) 22.0 ⫾ 2.16 (10) 48.0 (1) 30.0 ⫾ 2.00 (6) 50.2 ⫾ 6.14 (5) 38.5 ⫾ 3.87 (4) 41.0 ⫾ 1.73 (3) 32.4 ⫾ 9.24 (204) 24.5 ⫾ 2.22 (10) 21.5 ⫾ 1.27 (10) 18.0 (1) 23.9 ⫾ 1.85 (10) 20.5 ⫾ 0.71 (2) 18.6 ⫾ 0.92 (8) 25.3 ⫾ 1.96 (23) 27.5 ⫾ 0.71 (2) 27.7 ⫾ 2.06 (7) 12.8 ⫾ 0.29 (4) 46.0 (1) 33.9 ⫾ 2.54 (7) 15.7 ⫾ 1.27 (2) 25.3 ⫾ 2.30 (6) 13.89 ⫾ 0.65 (9) 26.8 ⫾ 2.59 (5) 21.3 ⫾ 0.96 (4) 22.7 ⫾ 1.03 (6) 15.8 ⫾ 2.22 (4) 29.8 ⫾ 2.59 (9) 15.0 ⫾ 1.41 (2) 22.8 ⫾ 1.60 (6) 27.0 ⫾ 2.16 (4) 13.5 ⫾ 0.87 (3) 29.8 ⫾ 5.48 (3) 38.9 ⫾ 4.23 (9) 15.0 (1) 26.0 ⫾ 1.41 (2) 30.8 ⫾ 1.48 (5) 14.7 ⫾ 0.82 (10) 20.4 ⫾ 2.22 (10) 35.0 (1) 18.4 ⫾ 1.50 (6) 35.2 ⫾ 4.44 (5) 28.8 ⫾ 2.63 (4) 24.3 ⫾ 3.06 (3) 24.0 ⫾ 6.91 (204) 1.39 ⫾ 0.097 (10) 1.29 ⫾ 0.057 (10) 1.39 (1) 1.38 ⫾ 0.129 (10) 1.25 ⫾ 0.077 (2) 1.31 ⫾ 0.091 (8) 1.39 ⫾ 0.084 (23) 1.22 ⫾ 0.109 (2) 1.19 ⫾ 0.082 (7) 1.35 ⫾ 0.008 (4) 1.63 (1) 1.29 ⫾ 0.154 (7) 1.62 ⫾ 0.113 (2) 1.54 ⫾ 0.118 (6) 1.73 ⫾ 0.068 (9) 1.16 ⫾ 0.066 (5) 1.40 ⫾ 0.123 (4) 1.19 ⫾ 0.070 (6) 1.52 ⫾ 0.148 (4) 1.28 ⫾ 0.062 (9) 1.50 ⫾ 0.095 (2) 1.27 ⫾ 0.088 (6) 1.28 ⫾ 0.065 (4) 1.70 ⫾ 0.126 (3) 1.54 ⫾ 0.083 (3) 1.29 ⫾ 0.131 (9) 1.60 (1) 1.25 ⫾ 0.014 (2) 1.21 ⫾ 0.076 (5) 1.30 ⫾ 0.086 (10) 1.08 ⫾ 0.103 (10) 1.37 (1) 1.64 ⫾ 0.160 (6) 1.43 ⫾ 0.064 (5) 1.34 ⫾ 0.036 (4) 1.70 ⫾ 0.209 (3) 1.36 ⫾ 0.182 (204) *Error ranges represent one standard deviation and numbers in parentheses are number of specimens per genus. of the estimate for cochlear impedance (T1: U ⫽ 12.5, P ⬍ 0.001; T2: U ⫽ 12.0, P ⬍ 0.001). The component of the ITR that is primarily responsible for this pattern is the ossicular lever arm ratio. The areal convergence ratios of anthropoids and strepsirrhines are not significantly different (U ⫽ 77; P ⫽ 0.874), but the difference in lever ratios is highly significant (U ⫽ 17; P ⬍ 0.001). It is noteworthy that lever ratio is significantly correlated (r ⫽ ⫺0.487; P ⫽ 0.012) with overall size across all primates, and anthropoids and strepsirrhines differ significantly in overall size. To determine whether the suborder difference in lever ratio persists when this size difference is taken into account, an ANCOVA was performed on lever ratio with size (GM) as a covariate. This ANCOVA confirmed significant differences between anthropoids and strepsirrhines in lever ratio independent of their size differences (F ⫽ 74.55; P ⬍ 0.001). To reveal whether the difference between anthropoids and strepsirrhines in lever ratios is due to differences in malleolar or incudal lever arm lengths, ANCOVAs were performed to identify suborder differences in lever arms with size (GM) as a covariate. This revealed no differences between suborders in incudal lever arm length, but significant differences in malleolar lever arm length (F ⫽ 23.06; P ⬍ 0.001). Thus, suborder differences in ITR are primarily driven by differences in length of the manubrium of the malleus (Fig. 9). Although areal convergence ratio data are not available for tarsiers, the ossicular lever arm ratio of 1.68 (derived from a single specimen) is clearly within the anthropoid range (1.58 –1.86) and outside the range for strepsirrhines (1.83–2.34). When suborder differences in T values are analyzed using a phylogenetically adjusted critical value of the F ratio (Fphylo ⫽ 32.96), the suborder differences are not significant despite the high F ratio (F ⫽ 19.90). In contrast, the ANCOVA testing for suborder differences in lever ratio while correcting for size remained significant (F ⫽ 74.55), even relative to a phylogenetically adjusted critical value of Fphylo ⫽ 33. ANCOVAs testing for suborder differences in malleus and incus lever arms with size as a covariate were not significant. Audiograms Lorisoid audiograms were compared with platyrrhine audiograms (Fig. 10A) since it was found that the mean body mass for the species in each group (608 and 656 g, respectively) is not significantly different (U ⫽ 4; P ⫽ 0.827) and the species in one group have equally sized AUDITORY DIVERSITY IN PRIMATES 1131 frequency sensitivity (U ⫽ 0; P ⫽ 0.016). However, the values for threshold of the primary peak are nonsignificant, despite there being over 6 dB difference between the means for the two species. A significant difference in highfrequency sensitivity was also detected (U ⫽ 0; P ⫽ 0.016). However, this difference is obviously related to the restricted high-frequency limit in Callithrix (Fig. 10A) and cannot be said to be characteristic of platyrrhine-lorisoid comparisons in general. Morphometrics vs. Audiometrics Fig. 7. A: Pinna shape index values (Psi) for families/subfamilies investigated in this study (Cebus and Saimiri plotted separately) showing that the major differences in outer ear shape occur between anthropoids and all other primates. Error bars represent two standard deviations. B: Estimated pinna shape index values for selected nodes on phylogenetic tree in Figure 6. Pinna shape index is plotted on the abscissa in the same scale as A. The ordinate is time (Myr), from 85 Myr at the bottom to the present at the top. Estimates are phylogenetically weighted means of the tip data calculated in PDTREE. Error bars represent standard error of the estimate. Lines connecting nodes are phylogenetic lines of descent. counterparts in the other. The most obvious difference between lorises and platyrrhines is in their low-frequency sensitivity where there is no overlap between the species from each group. Both low-frequency cutoff and low area are significantly different (U ⫽ 0; P ⱕ 0.05) with platyrrhines showing an average increase in sensitivity of around 15 dB in the lower range. Full area and middle area as well as overall range were also found to be greater in platyrrhines compared with lorisoids (all three: U ⫽ 0; P ⫽ 0.05). These differences seem to be related to the accentuated secondary peak evident in platyrrhine audiograms (local maxima around 2 kHz; Fig. 10A), but only minimally evident in lorisoid audiograms. Threshold of the primary peak is the remaining audiometric variable that shows significant differences (U ⫽ 0; P ⫽ 0.05) and indicates that platyrrhines are about 9 dB more sensitive at the frequency of greatest sensitivity than are lorisoids. When phylogenetically adjusted critical values of F ratios are calculated, only the suborder differences in low area remained significant (F ⫽ 30.40; Fphylo ⫽ 25.70). To assess the impact of intraspecific variation on comparisons between audiograms, the audiograms for Callithrix jacchus and Galago senegalensis were compared and are shown in Figure 10B. Although the galago data are less complete and appear more variable than the marmoset data, Callithrix still shows significantly better low- The results of the correlation analyses using both the original (tip) data and independent contrasts are given in Table 4. Using tip data, Psi shows a positive correlation with frequency of the primary peak (best Hz; r ⫽ 0.604; P ⫽ 0.042) but a negative correlation with low-frequency sensitivity (low cutoff and area), total range, and total audible area (r ⫽ 0.677, P ⫽ 0.023; r ⫽ ⫺0.736, P ⫽ 0.012; r ⫽ ⫺0.631, P ⫽ 0.034; r ⫽ ⫺0.759, P ⫽ 0.009). Note that an increasing value for low cutoff corresponds to decreased low-frequency sensitivity. Hence, a negative relationship is interpreted from positive correlation coefficient values. The correlation analyses between Psi and audiometric variables using independent contrasts produced rather different results from those obtained using tip data (Table 4). None of the correlations between pinna shape index and the audiometric variables are significant at P ⬍ 0.05 or 0.10. The highest correlations are between Psi and total range (r ⫽ ⫺0.476; P ⫽ 0.117) and total area (r ⫽ ⫺0.460; P ⫽ 0.126), although both are just above the critical P value. Using tip data, there is a weak negative correlation between T1 and low area (r ⫽ ⫺0.626; P ⫽ 0.092). Across all primates, there is a positive correlation between T1 values and the thresholds of the primary peak (best dB; r ⫽ 0.636; P ⫽ 0.087), i.e., as transmission percentage goes up, so does the loudness level (dB) required to hear a sound at the most sensitive frequency. This result is paradoxical because animals should be more sensitive as their transmission percentage increases, not less. This result is attributable to suborder differences. Strepsirrhines have lower sensitivity compared with anthropoids, but highly elevated T1 values (Fig. 11), producing a positive correlation between T1 and sensitivity across all primates. However, correlation coefficients calculated within suborders reveal the expected negative correlation between T1 and threshold of the primary peak (anthropoids, contrasts r ⫽ ⫺0.765; strepsirrhines contrast r ⫽ ⫺0.262). As with the correlations between Psi and audiometric variables, none of the correlations between T1 and audiometric variables were significant when contrasts were used. The lowest P value was produced by the negative correlation between T1 and frequency of the primary peak (r ⫽ ⫺0.614; P ⫽ 0.135). DISCUSSION Phylogenetic Effects on Primate Hearing This study identified significant distinctions in outer ear shape (Psi), theoretical middle ear transmission properties (T), and hearing sensitivity (areal measures, low-frequency cutoff, threshold of the primary peak) at the subordinal level. These distinctions were highly significant by the standards of traditional statistics, but many of these TABLE 3. Measurements taken on middle ear structures used to calculate ITR and T values* Genus Alouatta Aotus Arctocebus Ateles Avahi Brachyteles Cacajao Callicebus Callimico Callithrix Cebuella Cebus Chiropotes Daubentonia Euoticus Galago Galagoides Hapalemur Indri Lagothrix Lemur Leontopithecus Lepilemur Macaca Microcebus Miopithecus Perodicticus Phaner Pithecia Propithecus Saguinus Saimiri Tarsius Varecia Total Tympanic memrbane area (mm2) 51.5 ⫾ 5.03 (25) 25.8 ⫾ 3.11 (24) 23.7 ⫾ 2.33 (2) 56.9 ⫾ 9.13 (9) 26.2 ⫾ 0.52 (3) 49.2 (1) 32.0 ⫾ 3.26 (22) 31.0 ⫾ 3.51 (17) 24.2 ⫾ 2.33 (4) 20.0 ⫾ 1.69 (20) 14.3 ⫾ 1.01 (4) 38.6 ⫾ 4.00 (57) 32.5 ⫾ 3.00 (19) 43.1 ⫾ 2.65 (5) 17.8 ⫾ 0.21 (2) 22.1 ⫾ 1.89 (18) 23.0 ⫾ 2.18 (13) 36.0 ⫾ 2.89 (5) 46.2 ⫾ 3.66 (13) 26.9 ⫾ 2.70 (26) 23.7 ⫾ 2.01 (10) 30.5 ⫾ 3.30 (8) 13.0 ⫾ 4.27 (8) 25.3 ⫾ 1.76 (17) Stapedial footplate area (mm2) Areal convergence ratio 1.53 ⫾ 0.14 (20) 0.77 ⫾ 0.08 (15) 0.78 ⫾ 0.11 (2) 1.57 ⫾ 0.25 (10) 0.82 (1) 1.43 (1) 1.08 ⫾ 0.01 (2) 0.90 ⫾ 0.11 (9) 0.59 ⫾ 0.09 (3) 0.55 ⫾ 0.04 (5) 0.40 ⫾ 0.02 (2) 1.06 ⫾ 0.14 (29) 0.97 ⫾ 0.13 (5) 1.32 ⫾ 0.17 (2) 22.21 22.11 20.05 23.92 21.41 22.71 19.56 22.73 27.07 24.00 23.60 24.04 22.11 21.55 0.53 ⫾ 0.06 (4) 27.52 1.05 ⫾ 0.18 (2) 1.67 ⫾ 0.18 (2) 0.67 (1) 0.65 ⫾ 0.08 (6) 0.73 ⫾ 0.15 (5) 1.17 (1) 0.37 ⫾ 0.11 (2) 22.63 18.26 26.50 24.06 27.58 23.19 38.8 ⫾ 4.50 (20) 29.5 ⫾ 2.83 (17) 20.4 ⫾ 1.16 (13) 20.5 ⫾ 2.10 (57) 0.73 ⫾ 0.06 (8) 0.54 (1) 0.99 ⫾ 0.12 (8) 0.97 ⫾ 0.11 (2) 0.45 ⫾ 0.05 (4) 0.62 ⫾ 0.06 (14) 22.87 25.87 20.07 29.92 21.82 30.7 ⫾ 2.25 (10) 30.4 ⫾ 10.44 (449) 0.95 ⫾ 0.37 (168) 21.12 (26) Malleus lever length (mm) Incus lever length (mm) Ossicular lever arm ratio 3.57 ⫾ 0.30 2.84 ⫾ 0.14 2.90 ⫾ 0.10 3.24 ⫾ 0.08 3.31 ⫾ 0.06 3.30 2.89 ⫾ 0.26 3.02 ⫾ 0.19 2.43 ⫾ 0.10 2.47 ⫾ 0.14 2.03 ⫾ 0.04 3.05 ⫾ 0.19 2.85 ⫾ 0.13 4.78 2.35 2.67 ⫾ 0.12 2.04 3.18 ⫾ 0.13 3.65 ⫾ 0.14 3.14 ⫾ 0.25 3.41 ⫾ 0.10 2.76 ⫾ 0.08 3.44 ⫾ 0.12 3.77 ⫾ 0.36 2.44 ⫾ 0.08 3.14 2.87 ⫾ 0.18 2.73 3.01 ⫾ 0.15 4.12 ⫾ 0.05 2.47 ⫾ 0.11 2.35 ⫾ 0.09 2.47 3.62 ⫾ 0.19 2.91 ⫾ 0.43 2.24 ⫾ 0.14 1.66 ⫾ 0.14 1.44 ⫾ 0.07 1.99 ⫾ 0.25 1.59 ⫾ 0.06 2.06 1.78 ⫾ 0.22 1.68 ⫾ 0.10 1.36 ⫾ 0.03 1.36 ⫾ 0.10 1.11 ⫾ 0.06 1.90 ⫾ 0.13 1.57 ⫾ 0.10 2.20 1.16 1.19 ⫾ 0.07 0.98 1.57 ⫾ 0.04 1.92 ⫾ 0.20 1.82 ⫾ 0.09 1.62 ⫾ 0.03 1.49 ⫾ 0.09 1.48 ⫾ 0.13 2.35 ⫾ 0.05 1.23 ⫾ 0.12 1.98 1.39 ⫾ 0.08 1.49 1.76 ⫾ 0.09 2.16 ⫾ 0.23 1.34 ⫾ 0.16 1.39 ⫾ 0.06 1.47 1.98 ⫾ 0.16 1.62 ⫾ 0.42 1.60 ⫾ 0.18 (6) 1.72 ⫾ 0.10 (23) 2.01 ⫾ 0.03 (2) 1.64 ⫾ 0.24 (2) 2.08 ⫾ 0.11 (2) 1.60 (1) 1.63 ⫾ 0.10 (12) 1.80 ⫾ 0.11 (10) 1.78 ⫾ 0.04 (2) 1.82 ⫾ 0.13 (25) 1.85 ⫾ 0.08 (3) 1.61 ⫾ 0.11 (37) 1.82 ⫾ 0.17 (4) 2.17 (1) 2.03 (1) 2.24 ⫾ 0.07 (18) 2.08 (1) 2.02 ⫾ 0.04 (3) 1.91 ⫾ 0.13 (2) 1.72 ⫾ 0.06 (4) 2.10 ⫾ 0.06 (5) 1.85 ⫾ 0.11 (7) 2.34 ⫾ 0.15 (5) 1.60 ⫾ 0.18 (3) 1.99 ⫾ 0.13 (5) 1.59 (1) 2.06 ⫾ 0.05 (8) 1.83 (1) 1.71 ⫾ 0.10 (7) 1.92 ⫾ 0.23 (2) 1.86 ⫾ 0.19 (4) 1.70 ⫾ 0.08 (8) 1.68 (1) 1.84 ⫾ 0.22 (7) 1.82 ⫾ 0.23 (223) T1 T2 0.0178 0.0152 0.0122 0.0157 0.0109 0.0170 0.0191 0.0136 0.0116 0.0125 0.0125 0.0160 0.0137 0.0097 83.3 88.3 94.1 87.0 96.4 84.9 80.7 91.6 95.2 93.4 93.4 86.3 91.1 98.1 57.1 63.0 71.6 61.8 75.7 58.8 54.4 67.4 73.4 70.5 70.7 61.1 67.0 80.0 0.0072 99.9 89.8 0.0121 0.0183 0.0085 0.0121 0.0066 94.1 82.5 99.6 94.1 99.7 71.9 56.0 84.7 71.8 92.0 0.0108 96.6 76.2 0.0102 97.6 78.1 0.0131 0.0135 0.0108 0.0158 92.0 91.6 96.5 87.0 68.7 67.5 76.0 61.5 0.0129 (26) 92.1 70.3 ITR *T1 is the percentage of transmission through the middle ear calculated using 5,600 dynes sec/cm3 as an estimate of the specific acoustic impedance of the inner ear and T2 is calculated using 11,200 dynes sec/cm3. Areal convergence ratio, ITR, T1, and T2 calculated using genus means. Two-thirds correction factor applied to tympanic membrane area before calculating areal convergence ratio. Error ranges represent one standard deviation and numbers in parentheses are number of specimens/taxa. AUDITORY DIVERSITY IN PRIMATES 1133 Fig. 8. Theoretical percentage of acoustic transmission through the middle ear (T) based on ITR values derived from areal convergence and ossicular lever arm ratios. There is a significant difference in T between anthropoids and strepsirrhines despite the intermediate values for Callitrichinae. Fig. 10. Audiograms of similarly sized lorisoids and platyrrhines. A: Significant differences were found between the means for these two groups in low-frequency sensitivity, audible area (except high area), range, and threshold of the primary peak. B: When intraspecific variation is considered, platyrrhines (Callithrix) still show better low-frequency sensitivity than lorisoids (Galago), but the other audiometric traits become nonsignificant. Although a difference in high-frequency sensitivity was detected, this appears related to the reduced high-frequency sensitivity of Callithrix and is not characteristic of platyrrhines in general. Fig. 9. Malleus-incus pairs of two similarly sized primates (body mass and geometric mean of skull dimensions), Varecia (left) and Cebus (right), illustrating the relatively longer manubrium of strepsirrhines compared with anthropoids. The relative lengths of the incudal lever arms were not found to differ between these two groups, suggesting that malleolar lever arms are ultimately responsible for the increased ITR and T values of strepsirrhines compared with anthropoids. differences became nonsignificant when the degrees of freedom were adjusted to account for phylogenetic nonindependence of the data. The loss of significance to these suborder differences is due to the marked elevation in the critical values of the F ratio generated by the Monte Carlo simulations. Strictly speaking, these results suggest that the differences between the extant members of the two suborders are not larger than those obtained in 95% of cases of simulated evolution of these traits on the tree in Figure 6. We investigated the sensitivity of these results to different assumptions of the simulations—initial or starting values, boundary conditions, etc.—and found them to be fairly robust. The critical values of the F ratio generated by the simulations (Fphylo) were always much higher than those obtained using traditional statistics. Why do the traditional statistics and the phylogenetically adjusted statistics give such different results? One explanation is that many of the suborder differences arose in the basal branches of the haplorhine (or anthropoid) and strepsirrhine clades, with little further change along descendent branches. This is shown in Figure 7B, where the evolution of pinna shape (Psi) is reconstructed at selected primate nodes. This figure illustrates how the distribution of pinna shapes in extant primates (summarized in the traditional box plot in Fig. 7A) may have come about. Little change in Psi is evident in the lineages leading to extant strepsirrhines and in many lineages leading to extant anthropoids. The notable exceptions are the stem lineages of extant anthropoids, leading from the haplorhine node (at 67.5 Myr) to the basal anthropoid node (50 Myr) and from there to the platyrrhine (25 Myr) and catarrhine (35 Myr) nodes. Most of the difference in Psi between anthropoids and other primates arose in these three basal branches, and the phylogenetically adjusted critical values of the F ratio reflect this. However, this example is noteworthy in that the anthropoid versus nonanthropoid difference remains significant even when phylogenetic topology is taken into account. Similar effects on critical values of the F ratio rendered differences in T between anthropoids and strepsirrhines nonsignificant. 1134 COLEMAN AND ROSS TABLE 4. Correlation coefficients (r) and probability values (P) for correlations between audiometric variables and Psi and T* Psi T Tips Best Hz Best dB Low cutoff High cutoff Low area Middle area High area Total area Total range Pics Tips Pics r P r P r P r P 0.604 0.451 0.677 0.219 ⫺0.736 ⫺0.814 0.055 ⫺0.759 ⫺0.631 0.04 0.11 0.02 0.29 0.01 ⬍ 0.01 0.44 ⬍ 0.01 0.03 0.357 0.013 ⫺0.139 ⫺0.196 ⫺0.194 ⫺0.391 ⫺0.321 ⫺0.460 ⫺0.476 0.19 0.49 0.37 0.32 0.32 0.17 0.22 0.13 0.12 0.242 0.636 0.595 0.500 ⫺0.626 ⫺0.436 0.124 ⫺0.554 ⫺0.509 0.32 0.09 0.11 0.16 0.09 0.19 0.41 0.13 0.15 ⫺0.614 0.147 0.130 0.106 ⫺0.108 0.092 ⫺0.388 ⫺0.349 ⫺0.437 0.14 0.41 0.42 0.43 0.43 0.44 0.26 0.28 0.23 *Results using tip data and standardized independent contrasts are given. Tips, r and P calculated using tip data; Pics, r and P calculated using standardized independent contrasts on tree with all branch lengths equal to 1.0. See Figure 5 for audiometric variable descriptions. Fig. 11. The relationship between the theoretical percentage of transmission through the middle ear and threshold of the primary peak taken from audiograms is in the opposite direction of that expected when considering primates as a group, but appears to agree with theoretical expectations when compared within anthropoids and strepsirrhines. When changes are concentrated in only two basal branches, similar results are more likely to be obtained in the simulations than if the changes accumulated in many branches. In effect, the highly inflated phylogenetically adjusted critical values of the F ratios reflect the possibility that the differences between extant members of haplorhines and strepsirrhines arose in the basal branches of the two clades and remained relatively unchanged since that time through either phylogenetic inertia or constraint. Another possibility is that natural selection has acted to maintain these basal differences along all descendent lineages, producing the suborder differences characterizing the extant representatives of the two suborders. For example, it is possible that activity cycle (i.e., diurnal or nocturnal activity) accounts for the observed difference in Psi, since all but one of the prosimians are noctur- nal (in this study) and all but one of the anthropoids are diurnal. ANOVA of pinna ratio by activity cycle (F ⫽ 27.27) is significant using a traditional critical value of the F ratio (P ⫽ 0.001; F ⫽ 13.3) and using the phylogenetically adjusted critical value of F (Fphylo ⫽ 27.30). This suggests that there are effects of activity pattern on pinna ratio independent of the effects of phylogenetic relatedness. Another possibility is that requirements for group communication and predator-prey interactions have influenced and maintained particular pinna morphologies. To test these hypotheses, the relationships between these morphological and audiometric variables and various ecological variables must be explored. The effects of phylogenetic relatedness on correlations between audiometric and morphological variables are more severe. None of the correlations between T or Psi and the audiometric variables were significant when phylogenetic independent contrasts were used. The reason for this is illustrated in Figure 12, a plot of total audible area of the audiogram against Psi. There is a significant negative correlation between these variables when tip data are used (r ⫽ ⫺0.760; P ⫽ 0.02), but this relationship is highly influenced by phylogeny and is not significant using phylogenetic contrasts (r ⫽ ⫺0.460; P ⫽ 0.21). All the strepsirrhines measured have high pinna shape indexes and low values of total area, whereas all the anthropoids have low pinna shape indexes and high values of total area. The use of independent contrasts takes these phylogenetic effects into account and, in this case, suggests that the relationship between these variables incorporates significant components of phylogenetic signal. Thus, the correlation might exist because these two variables are correlated with any of the many other features distinguishing anthropoids and strepsirrhines, or because of phylogenetic inertia. The data do not currently allow these possibilities to be distinguished. The strong phyogenetic effects documented by this study suggest that test of hypotheses regarding evolution of primate hearing must take phyogenetic relationships into consideration. Morphological Effects on Primate Hearing The most surprising finding from this study was the apparent lack of association between theoretical percentages of acoustic transmission (based on ITR values) and AUDITORY DIVERSITY IN PRIMATES Fig. 12. Scatterplot illustrating the relationship between total audible area and pinna shape index (Psi). Although there appears to be a significant correlation between these two variables, this relationship becomes nonsignificant when evaluated using phylogenetic independent contrasts. measures of sensitivity taken from audiograms. When the relationship between T and the threshold of the primary peak is examined within suborders, the general pattern agrees with theoretical expectations, i.e., increased acoustic transmission is associated with increased auditory sensitivity. However, when comparisons are made across primates as a whole, suborder differences in T values confound this relationship. Compared with anthropoids, strepsirrhines have enhanced T values but reduced auditory sensitivity (Fig. 11). The component of the ITR primarily responsible for this suborder difference in T values is the ossicular lever arm ratio. It may be that the increased lever arm advantage of strepsirrhines provides an increase in impedance matching performance but that this advantage is offset by other, yet to be identified acoustical factors. One possibility is that strepsirrhines have considerably higher cochlear impedance as a group compared with haplorhines and that using a constant value for all primates obscures the actual (versus theoretical) percentage of acoustic energy that makes it into the inner ear. It is also possible the mechanical impedance of the middle ear, determined by the mass, stiffness, and frictional resistance of the middle ear components, differs between the two groups. Until these discrepancies are more clearly understood, this finding should serve as a cautionary note to researchers who use ITRs as direct measures of hearing function in comparative studies (Lay, 1972; Hunt and Korth, 1980; Masali et al., 1992). The advantage of using a constant value for the specific acoustic impedance of the cochlea is that it allows the efficiency of the middle ear to be evaluated in isolation without influence from other auditory components. Furthermore, it permits these data to be compared directly with other studies that used a similar approach. Within primates, the T1 values for most ape taxa (adjusted using a 2/3 correction factor for the effective area of the tympanic membrane) (Masali et al., 1992) are similar to other 1135 haplorhines: Pongo ⫽ 80%, Pan ⫽ 84%, and Gorilla ⫽ 92%. Humans are the standout among the primates that have been tested with T1 values around 46%. Compared with other mammalian orders, primates have fairly typical middle ear transmission properties despite the seemingly high values (T1 range ⫽ 81% to 100%). The New World heteromyid desert rodents were found to have T values similar to primates falling into two primary groups: the genera Dipodomys, Microdidodops, and Perognathus range from 94% to 100, while species of Liomys have somewhat lower values, between 78% to 80 (Webster and Webster, 1975). Similar to primates, differences in the lever ratio (particularly the malleus) appear to be the primary mechanism causing differences in ITR and ultimately T values (Webster and Webster, 1975). Among archontan taxa, dermopterans (Cynocephalus volans) have an average T value of 91% but scandentians (Tupaia glis) appear less efficient with a value around 70% (Hunt and Korth, 1980). It is unlikely that the differences in pinna morphology completely account for the total increase in low-frequency sensitivity witnessed in platyrrhines compared with lorisoids (and by extension anthropoids and prosimians). Shaw (1974) and others have found that the total contribution of all three components of the human outer ear (pinna, concha, and ear canal) adds up to approximately a 20 dB acoustic gain between 1 and 8 kHz (centered at 2.7 kHz) with considerably less gain at higher and lower frequencies. Similar values have been reported for cats (Wiener et al., 1966) and rabbits (Fattu, 1969), while guinea pigs show an increase of just over 10 dB (Sinyor and Laszlo, 1973). However, the amount of acoustic amplification provided by the pinna alone is only a modest 5 dB in humans and between 5 and 10 dB in cats, rabbits, and guinea pigs (at the best frequency). All of these studies indicate that the ear canal and, to a lesser degree, the concha contribute the majority of pressure gain provided by the outer ear. It should be kept in mind that these pressure increases are maximal in the mid frequency range (as defined in this study) and provide significantly less amplification at lower frequencies. Furthermore, an acoustic gain of 5 db, for example, provided by a single component of the auditory system does not necessarily result in a 5 dB increase in overall threshold values in audiograms due to the interaction of the different auditory components (middle ear, inner ear, as well as different parts of the outer ear). For example, the outer ear may produce a resonance (acoustic gain) in a frequency range that is countered by an antiresonance (acoustic loss) produced by the middle ear, resulting in little or no net amplification. As pointed out above, the secondary peak seen in platyrrhine audiograms seems to be the proximate mechanism responsible for the increase in low-frequency sensitivity. This secondary peak may be the result of the different middle ear cavity configurations displayed by anthropoids and strepsirrhines. Acoustic theory suggests that dual- or multichambered middle ear cavities will result in more than one resonant frequency with an increase in impedance at the transition between the resonant frequencies (Dallos, 1973; Moore, 1981), resulting in enhanced auditory reception of both high- and low-frequency sounds (Moore, 1981; Lombard and Hetherington, 1993). This is precisely the pattern illustrated in anthropoid audiograms: two resonant peaks separated by a trough accom- 1136 COLEMAN AND ROSS panied by increased low-frequency sensitivity without a significant loss of high-frequency sensitivity. However, the problem with this generalization is that while lorisoids resemble anthropoids in having additional pneumatic spaces off of the tympanic cavity (e.g., medial accessory cavity sensu, MacPhee, 1981) that may function as acoustically significant accessory chambers, lorisoids resemble lemuroids and differ from anthropoids in their relatively diminished low-frequency sensitivity. Research is currently being conducted to investigate the influence of primate middle ear cavity volume and different cavity configuration patterns on hearing performance. Regardless of the specific auditory components responsible for augmenting particular aspects of hearing sensitivity, the finding that platyrrhines have significantly better low-frequency sensitivity than like-sized lorisoids could lead to a new understanding of how these groups have adapted to their respective ecologies and habitats. An increase in low-frequency sensitivity could result in a considerable expansion of the distance over which primates can adequately communicate. Brown and Waser (1984) found that a heightened sensitivity of around 10 dB in the low-frequency range of blue monkeys (Cercopithecus mitis) results in a four-fold increase in the audible distance of their low-frequency boom calls. The average difference in platyrrhine and lorisoid hearing between 250 and 1,000 Hz is 14.6 dB, suggesting that New World monkeys have the potential to benefit from a considerable increase in the audible distance of long calls, although the exact propagation distances are related to particular aspects of different environments (Waser and Brown, 1986). This study reveals that reshaping the outer ears may be one way that anthropoids have increased their low-frequency hearing sensitivity. Although the middle ear has also changed in morphology and presumably functionality, much future research is needed to understand fully the exact influence that these changes have on auditory processing and what impact they had on the evolution of anthropoids and primates in general. ACKNOWLEDGMENTS The authors thank Ted Garland and Bill Jungers for advice and assistance with statistics. They greatly appreciate the attention of John J. Rosowski, who provided helpful discussion and detailed comments on the manuscript. Tim Smith also provided useful comments on the manuscript. Outer ear illustrations were drawn by Colleen Lodge. LITERATURE CITED Beecher MD. 1974a. Pure-tone thresholds of the squirrel monkey (Saimiri sciureus). J Acoust Soc Am 55:196 –198. Beecher MD. 1974b. Hearing in the owl monkey (Aotus trivirgatus): I, auditory sensitivity. J Comp Phys Psych 86:898 –901. Brown CH, Waser PM. 1984. Hearing and communication in blue monkeys (Cercopithecus mitis). Anim Behav 32:66 –75. Burrows AM, Smith, TD. 2003. Muscles of facial expression in Otolemur, with a comparison to Lemuroidea. Anat Rec 274A:827– 836. Cartmill M. 1975. Strepsirhine basicranial structures and the affinities of the Cheirogaleidae. In: Luckett WP, Szalay FS, editors. Phylogeny of the primates: a multidisciplinary approach. New York: Plenum Press. p 313–354. Coleman MN. 2003. An investigation of scaling relationships in visual and masticatory systems of New World primates. Am J Phys Anthropol 36(Suppl):78. Dallos P. 1973. The auditory periphery: biophysics and physiology. New York: Academic Press. Fattu JM. 1969. Acoustic orientation by the rabbit pinna and external auditory meatus. J Acoust Soc Am 16:124. Felsenstein J. 1985. Phylogenies and the comparative method. Am Nat 125:1–15. Fleagle JG. 1999. Primate adaptation and evolution. San Diego, CA: Academic Press. Garland T, Dickerman AW, Janis CM, Jones JA. 1993. Phylogenetic analysis of covariance by computer simulations. Syst Biol 42:165– 292. Gillette RG, Brown R, Herman P, Vernan S, Vernon J. 1973. The auditory sensitivity of the lemur. Am J Phys Anthropol 38:365–370. Heffner HE, Ravizza RJ, Masterton B. 1969. Hearing in primitive mammals: IV, bushbaby (Galago senegalensis). J Aud Res 9:19 –23. Heffner HE, Masterton B. 1970. Hearing in primitive mammals: slow loris (Nycticebus coucang) and potto (Perodicticus potto). J Comp Phys Psych 71:175–182. Hemilä S, Nummela S, Reuter T. 1995. What middle ear parameters tell about impedance matching and high frequency hearing. Hear Res 85:31– 44. Hershkovitz P. 1977. Living New World monkeys (Platyrrhini), vol. 1. Chicago: University of Chicago Press. Hienz RD, Turkkan JS, Harris A. 1982. Pure tone thresholds in the yellow baboon (Papio cynocephalus). Hear Res 8:71–75. Hill WCO. 1955. Primates, comparative anatomy and taxonomy: II, haplorhini, tarsioidea. Edinburgh: Edinburgh University Press. Huber E. 1930. Evolution of facial musculature and cutaneous field of trigeminus. Quart Rev Biol 5:133–188. Hunt R, Korth W. 1980. The auditory region of Dermoptera: morphology and function relative to other living mammals. J Morph 164: 167–211. Jackson LL, Heffner RS, Heffner HE. 1999. Free-field audiogram of the Japanese macaque (Macaca fuscata). J Acoust Soc Am 106: 3017–3023. Kojima S. 1990. Comparison of auditory functions in the chimpanzee and human. Folia Primatol 55:62–72. Lay DM. 1972. The anatomy, physiology, functional significance and evolution of specialized hearing organs of gerbilline rodents. J Morp 138:41–120. Lightoller GS. 1934. The facial musculature of some lesser primates and a Tupia: a comparative study. Proc Zool Soc 20:259 –309. Lombard RE, Hetherington T. 1993. Structural basis of hearing and sound transmission. In: Hanken J, Hall BK, editors. The skull, vol. 3, functional and evolutionary mechanics. Chicago: University of Chicago Press. p 241–302. MacPhee RDE. 1981. Auditory regions of primates and eutherian insectivores. In: Szalay FS, editor. Contributions to primatology, vol. 18. Basel: S. Karger. p 1–282. MacPhee RDE, Cartmill M. 1986. Basicranial structures and primate systematics. In: Swindler DP, Erwin J, editors. Comparative primate biology, vol. 1, systematics, evolution, and anatomy. New York: Alan R. Liss. p 219 –275. Masali M, Tarli SB, Maffei M. 1992. Auditory ossicles and the evolution of the primate ear: a biomechanical approach. In: Wind J, Chiarelli B, Bichakjian B, Nocentini A, Jonker A, editors. Language origin: a multidisciplinary approach. Dordrecht: Kluwer Academic. p 67– 86. Masterson B, Heffner H, Ravizza R. 1969. The evolution of human hearing. J Acoust Soc Am 45:966 –985. Moggi-Cecchi J, Collard M. 2002. A fossil stapes from Sterkfontein, South Africa, and the hearing capabilities of early hominids. J Hum Ev 42:259 –265. Moore WJ. 1981. The mammalian skull. Cambridge: Cambridge University Press. Olsen EC, Miller RL. 1958. Morphological integration. Chicago: University of Chicago Press. Owren MJ, Hopp SL, Sinnott JM, Peterson MR. 1988. Absolute auditory thresholds in three Old World monkey species (Cercopithecus aethiops, C. neglectus, Macaca fuscata) and humans (Homo sapiens). J Comp Psych 102:99 –107. AUDITORY DIVERSITY IN PRIMATES Packer DJ, Sarmiento EE. 1984. External and middle ear characteristics of primates, with reference to tarsier-anthropoid affinities. Am Mus Nov 2787:1–23. Pfingst BE, Laycock J, Flammino F, Lonsbury-Martin B, Martin G. 1978. Pure tone thresholds for the Rhesus monkey. Hear Res 1:43– 47. Rosowski JJ, Graybeal A. 1991. What did Morganucodon hear? Zool J Linn Soc 101:131–168. Rosowski JJ. 1994. Outer and middle ears. In: Fay RR. Popper AN, editors. Comparative hearing: mammals. New York: Springer Verlag. p 172–247. Ross CF, Henneberg M, Ravosa MJ, Richard S. 2004. Curvilinear, geometric and phylogenetic modeling of basicranial flexion: is it adaptive, is it constrained? J Hum Evol 46:185–213. Schultz AH. 1969. The life of primates. New York: Universe Books. Schultz AH. 1973. The skeleton of the Hylobatidae and other observations on their morphology. In: Rumbaugh DM, editor. Gibbon and siamang, vol. 2, anatomy, dentition, taxonomy, molecular evolution and behavior. Basel: S. Karger. p 1–54. Seiden HR. 1957. Auditory acuity of the marmoset monkey (Hapale jacchus). PhD dissertation. Princeton, NJ: Department of Psychology, Princeton University. Shaw EAG. 1974. The external ear. In: Keidel WD, Neff WD, editors. Handbook of sensory physiology, vol. V(1). Berlin: Springer Verlag. Sinyor A, Laszlo CA. 1973. Acoustic behavior of the outer ear of the guinea and the influence of the middle ear. J Acoust Soc Am 54: 916 –921. 1137 Smith RJ, Jungers WL. 1997. Body mass in comparative primatology. J Hum Ev 32:523–559. Spencer MA, Spencer GS. 1995. Technical note: video-based threedimensional morphomethrics. Am J Phys Anth 96:443– 453. Stebbins WC, Green S, Miller FL. 1966. Auditory sensitivity of the monkey. Science 153:1646 –1647. Stebbins WC. 1975. Hearing in anthropoid primates: a behavioral analysis. In: Tower DB, editor. The nervous system, vol. 3, human communication and its disorders. New York: Raven Press. p 113– 124. Waser PM, Brown CH. 1986. Habitat acoustics and primate communication. Am J Primatol 10:135–154. Webster DB, Webster M. 1975. Auditory systems of Heteromyidae: functional morphology and evolution of the middle ear. Morphology 146:343–376. Wiener FM, Pfeiffer RR, Backus ASN. 1966. On the sound pressure transformation by the head and auditory meatus of the cat. Acta Otolaryn 61:255–269. Yoder AD, Yang Z. 2004. Divergence dates for malagasy lemurs estimated from multiple gene loci: geological and evolutionary context. Mol Ecol 13:757–773. Zwislocki J. 1965. Analysis of some auditory characteristics. In: Luce RD, Bush RR, Galanter E, editors. Handbook of mathematical psychology. New York: John Wiley and Sons. p 1–98 Zwislocki J. 1975. The role of the external and middle ear in sound transmission. In: Tower DB, editor. The nervous system, vol. 3, human communication and its disorders. New York: Raven Press. p 45–55.