Cuvier's beaked whale (Ziphius cavirostris) head tissues:physical properties and CT imaging

Cuvier's beaked whale (Ziphius cavirostris) is the most abundant species of the Ziphiidae family of toothed whales(Heyning, 1989b), yet much about it remains unknown. Studies of live Z. cavirostris are rare, as they are visually inconspicuous at sea due to deep, long-duration dives and short surface times (Heyning,1989b; Barlow et al.,1997; Baird et al.,2004). Cuvier's beaked whales typically inhabit deep, offshore waters exhibiting steep slope features such as submarine canyons, oceanic islands, the continental shelf edge, and enclosed seas(Boutiba, 1994; D'Amico et al., 2003; Forney et al., 1995; Heyning, 1989b; Houston, 1991; Marini et al., 1996; Waring et al., 2001). Similar to other toothed whales, echolocation is important for Cuvier's beaked whales when navigating these features and when foraging for deep-sea and benthic squid, fish and crustaceans in the mesopelagic zone(Blanco and Raga, 2000; Fiscus, 1997; Johnson et al., 2004; MacLeod et al., 2003; Santos et al., 2001). Recently, the use of high-intensity anthropogenic sounds, particularly mid-frequency (2–4 kHz) sonar, has been associated with mass strandings of Cuvier's beaked whales (Evans and England, 2001; Frantzis,1998; Simmonds and Lopez-Jurado, 1991). Physiological mechanisms including rectified diffusion of gas bubbles in tissues (Crum and Mao, 1996; Houser et al.,2001), gas emboli formation in body fats(Jepson et al., 2003;Fernandez et al., 2004) or bubble or lung resonance(Finneran, 2003) may explain the observed hemorrhaging in the lungs, acoustic pathways, brain, spinal cord,kidneys and eyes, congestion and bubbles in the brain, and embolisms in body fats in stranded animals (Evans and England, 2001; Fernandez et al., 2004). The role that sound exposure played in these strandings and subsequent deaths remains controversial. Models of sound propagation within these beaked whales may provide insight into questions about their response to sound exposure,echolocation production and sound reception.

Sound production systems in beaked whales may be similar to those found in other odontocetes. Comparative anatomic studies highlight homologous structures responsible for sound generation associated with echolocation,which are unique among mammals (Au,1993; Cranford et al.,1996; Heyning,1989a; Heyning and Mead,1990; Lawrence and Schevill,1956; Mead, 1975). Convincing evidence suggests that echolocation clicks are produced at the phonic lips [a structural complex previously known as the monkey lips/dorsal bursae (MLDB) region; Cranford et al.,1996; Aroyan et al.,1992] and then progressively focused by the skull, air spaces,connective tissue sheaths and gradients within the melon, before propagating into the external environment. The melon, a complex connective tissue and lipid structure located in the forehead region of all odontocetes(Cranford et al., 1996), is likely to be important in the sound production system of Z. cavirostris (Fig. 1).

Morphological studies of homology across odontocetes suggest that the sound reception system in beaked whales is similar to the system found in better-studied odontocete species (Au,1993). The odontocete sound reception pathway involves transmission through bone and lipids of the lower jaw, rather than through the tympanic membrane, as is typical in terrestrial mammals. Sound waves are directed towards the inner ear complex through a system that includes fat bodies lining the interior and exterior mandibles, thin translucent regions of bones in the posterior portion of the lower jaws, and air sinuses(Aroyan, 2001; Brill et al., 1988; Figs 1, 2). Norris (1968) coined the term `acoustic fats' to describe these mandibular fat bodies, as well as the fats in the melon, due to their proposed role in sound propagation. The air spaces are excellent acoustic reflectors that may function to shield the ear complex from internally produced sounds and isolate them acoustically to facilitate directional hearing (Norris,1964).

Sound propagation modeling has been used only recently to investigate sound pathways in marine mammals. Aroyan et al.(1992) and Aroyan(2001) explored sound emission and reception in a common dolphin, Delphinus delphis, by combining three-dimensional acoustic propagation models with topographical X-ray computed tomography (CT) data. While these models show great promise for understanding sound propagation in marine mammals, they only used simplified estimates and assumptions of tissue properties, such as sound speed and density, and they do not consider the elastic properties of the tissues(Aroyan et al., 1992; Aroyan, 2001). The inclusion of measured tissue properties in sound propagation models will provide a more accurate representation of the complexities of sound emission and reception in these marine mammals.

Odontocete melon and mandibular fat bodies exhibit structural and chemical complexity that could have important implications for sound emission and reception. Structurally, the melon is comprised of gradations of tissues, with a central core of acoustic lipids grading into complex musculature, dense connective tissue (theca) and blubber, with notable topographical asymmetry(Heyning, 1989a; Cranford et al., 1996; Fig. 1). Studies of the chemical composition of melon acoustic lipids and mandibular fat bodies suggest complex chemical topography associated with acoustic functionality(Blomberg and Lindholm, 1976; Koopman et al., 2003; Litchfield et al., 1973; Varanasi et al., 1982). Sound speed in melon lipids has been shown to vary with chemical composition and topography in the melon (Blomberg and Jensen, 1976; Blomberg and Lindhom, 1976; Flewellen and Morris, 1978; Goold et al.,1996; Goold and Clarke,2000; Litchfield et al., 1973, 1979; Norris and Harvey, 1974; Varanasi et al., 1982). Sound speed in the tissues surrounding the melon has not been studied. Understanding structural and compositional heterogeneity, and the resulting tissue physical properties, is invaluable for developing a good representative model of the acoustic pathways in the head of an odontocete whale.

To build representative acoustic-structural models of a Cuvier's beaked whale, high-resolution sound speed, density and elasticity measurements are needed; however, no data exist in the literature for this species. Fortunately, in 2002, we obtained a freshly stranded neonate Cuvier's beaked whale specimen for study. From the head of this specimen, we present approximately 200 measurements for tissue sound speed, density, CT scan Hounsfield unit (HU) values and elastic modulus that provide a basis for which to build an acoustic-structural model. We show that our methods provide comparable results to those from aquatic and terrestrial mammals. We show that sound speed values and topographical variations in a neonate Cuvier's beaked whale melon are similar to those in more-studied odontocete species. Finally,we provide a predictable relationship between the Hounsfield units from CT scans and measured tissue sound speed and density values.

For CT imaging, dissection and tissue measurements, we obtained a carcass of a Cuvier's beaked whale, Ziphius cavirostris Cuvier (NMFS field number KXD0019). The 314.9 cm-long female neonate carcass, found on 7 January 2002 at Camp Pendleton in San Diego, CA, USA was determined by NMFS personnel to be freshly dead, with little evidence of decomposition. The animal was estimated to be about one month old at the time of death. Immediately after the body was recovered from the beach, it was completely frozen to–20°C. In April 2002, the frozen specimen was cut across the long axis of the body into four manageable segments using an industrial dual column band saw. The segments corresponded roughly to the head, thorax, abdomen and tail. The segments were individually thawed, CT scanned, dissected and cut into smaller pieces for further analysis. This report presents tissue properties for the head only; results from remaining segments and a more comprehensive list of measurement values for all segments can be found in Soldevilla et al. (2004).

The head was thawed in a water bath for 48 h to ensure complete, consistent thawing and then subjected to CT scanning using a GE Light Speed Plus CT scanner. Axial scans, 5 mm thick, were collected every 5 mm, to ensure the entire specimen was scanned. The scanner used a 500 mm-diameter field of view scan region. Images were collected with a 140 kV×320 mA power setting. Image data were recorded directly to hard drive.

We analyzed the CT scan data after the dissection to determine HU values for the volumes representing the tissue samples used for physical property measurements. We used E-film, a specialized software package (Merge-efilm,Milwaukee, WI, USA), to convert CT units of electron density into calibrated HUs. HUs are scaled from –1000 to >1000, where air is at –1000,room temperature water is at 0, and hard bone can be found at >1000(Robb, 1998). Mammalian soft tissues generally range between –100 and 100, with fatty tissues at the low end and denser connective tissues at the high end(Duck, 1990). HUs are standardized by defining water as zero and are calibrated such that one unit is equal to the change in the density of 1 cm³ of water raised 1°C at standard temperature and pressure. We referred to photos taken during the dissection to locate the position of dissected tissue samples in the CT scan images. We chose the CT image that corresponded to the center of each dissection slice by referring to landmarks such as the skull, eyes and ears. Samples were picked by visually correlating the CT image with the photo of each slice. HU values within a 50 mm³ region were used to calculate a mean HU and standard deviation at the approximate center of each sample.

A dissection was performed after the CT scan, with extra attention and finer scale sampling in the forehead and mandibular regions (see Table 1 for sample descriptions). The forehead was cut transversely into 2 cm-thick slices for a total of 15 slices (Fig. 3A,B). Each slice was further cut in a grid-like fashion into ∼2 cm cubes(Fig. 3C) to provide high-resolution topographical data. The left mandibular region was also cut transversely into 2 cm-thick slices for a total of 12 slices. For each slice,one sample was taken from the blubber and the exterior mandibular fat, and 1–5 samples were taken from the interior mandibular fat. Each cubic sample was numbered, notched on the anterior right dorsal corner, and stained on the anterior side with Basic Fuchsin(Lillie, 1977) to provide sample location and orientation reference during post-dissection evaluation. The dissection was recorded photographically with a 5 megapixel digital color camera (Sony Cyber-shot DSC-707).

Tissue samples were visually categorized by color, texture and location within the head into the following broad categories: blubber, acoustic fats(including melon and mandibular fat), muscle, and connective tissue. All samples containing a mixture of these categories were discarded from further analysis to provide homogeneous tissue samples for physical property measurements. Note that the tissue samples were categorized subjectively and,while more accurate descriptions of tissue could be determined by histological analysis, such an investigation was not within the scope of our analysis.

Sound velocity was measured in each tissue sample using a Krautkramer Branson USD10 Ultrasonic Digital Flaw Detector and two Krautkramer Branson longitudinal acoustic transducers (Alpha series, 10 MHz, 0.25 mm;Krautkramer,General Electric Inspection Technologies, Hverth, Germany) attached to digital calipers (model no. CD-8″CS, Mitutoyo America Corp., Aurora, IL, USA). The Krautkramer velocimeter system measured the transmission time of 10 MHz broadband pulses through the various tissue samples. The calipers were used to measure the sample thickness to the nearest millimeter, and velocity was calculated by dividing the thickness by the travel time. Prior to measuring the samples, the velocimeter was calibrated using distilled water at room temperature (22.5°C), assuming a sound speed of 1490 m s^–1 (Chen and Millero,1977). Sound velocity was measured along three directions of each sample cube (anterior–posterior, dorsal–ventral, lateral) to test for anisotropy. Temperatures of the samples were recorded for later temperature–sound velocity normalization. Ultrasonic attenuation measurements were also made. Results of sound velocity anisotropy and ultrasonic attenuation are reported by Soldevilla et al.(2004). Sound velocity measurements were made at ultrasonic (10 MHz) frequencies, because short wavelengths are necessary to measure small tissue samples, and dispersion effects are expected to be small (see below).

Temperature affects sound velocity in tissues. We were interested in estimating in vivo (37°C) sound velocity rather than room temperature (21°C) sound velocity so that future models can be representative of a living animal. A sample of each type of tissue was placed in consecutive water baths at temperatures ranging from 10 to 37°C until the sample reached equilibrium, and its sound velocity was measured for each temperature.

Longitudinal compression tests of each sample were performed to measure the elastic modulus. Tissue materials exhibit a nonlinear relationship between stress and strain. We determined the elastic modulus as the tangent to the nonlinear stress–strain curve at different stresses. A Mini Bionix 858(MTS Systems Corp., Eden Prairie, MN, USA) compressed each sample, while force and displacement data were recorded to a computer running TestStar (MTS Systems Corp.). The instrument was programmed to complete 10 low-stress cycles at 2 cycles s^–1 with a 0.08 mm displacement, followed by one high-stress cycle at 0.5 cycles s^–1 with a 2 mm displacement. Although these displacements may be large by acoustic standards, they enabled us to examine the nonlinearities in the tissues. The calipers that were used for the sound velocity measurements were also used to measure initial length of the tissue sample cube (i.e. initial distance between the compressing plates) and the area in contact with the plates. Each sample was placed in the instrument, and the hydraulic arm was engaged until the plates contacted the sample. If there was any initial force on the sample, it was recorded for use in later calculations, and the displacement gauge was zeroed prior to cycle loading the sample. We recorded force and displacement data at 100 samples s^–1. A customized MATLAB program provided stress versus strain plots for each sample. Hysteresis, a measure of the relative energy loss during elastic deformation, was calculated as the area between the loading curve and the recovery curve of the last low-stress cycle normalized by the total energy in the loading curve.

Density was calculated using Archimedes' principle. We measured volume by immersing each sample in distilled water and weighing the displaced water. Water temperature was monitored to allow accurate calculations of density from water volume measurements. Wet mass was determined by direct measurement with a Mettler PM 460 Delta Range mass balance (Mettler-Toledo Inc., Columbus, OH,USA). Density was calculated by dividing mass by volume.

Sound velocity measurements were averaged across orientations, resulting in non-directional sound speed. We performed one-factor analyses of variance(ANOVAs) to compare values among tissue types for sound speed, sound attenuation, density, HU and hysteresis, followed by two-factor ANOVAs to compare values between locations (forehead or mandible) for blubber and acoustic fats (Zar, 1999). ANOVAs were followed up with a Scheffe post-hoc analysis(Zar, 1999). All ANOVAs were performed using SYSTAT (Systat Software Inc., Point Richmond, CA, USA). Student's t-tests were run to compare tissue type values with those of seawater at 15°C, 1 atm. (=101 kPa) and 35‰ salinity. Linear regressions and a principal components analysis (PCA) were run to investigate the relationships between the various measures. First, we averaged the values from the three orientations for sound velocity and elasticity. This enabled us to compare these measures to the density and HU measures that only had one value per sample. We performed linear regressions to determine the relationship between HUs and each of the other measures. We used PCA to determine the independence of the four physical property variables and to isolate the most important modes of variability in the data. PCA seeks linear combinations of variables that maximally explain the variance in the data(Jackson, 1991). The variance of a large data set is concentrated into a small number of physically interpretable patterns of variability – the principal components. Patterns among the physical property variables are illustrated by loadings on the principal components (Jackson,1991).

Sound speed values of Ziphius cavirostris head tissues ranged between 1331 and 1708 m s^–1. Sound speed values corrected to 37°C were significantly different between all tissue types(P≪0.001), with connective tissue exhibiting the highest sound speeds, followed by muscle, blubber and acoustic fats(Fig. 5; Table 3). Fatty tissues in the forehead region had significantly higher sound speeds than fatty tissues in the mandibular region (P≪0.001). The mean sound speed of seawater is 1507 m s^–1 at 15°C, 101 kPa. and a salinity of 35‰ (Chen and Millero,1977). Acoustic fats and blubber exhibit sound speeds significantly lower than those of seawater (P<0.05). Muscle tissues exhibit sound speeds that are not statistically different from seawater, while connective tissue sound speeds were significantly higher(P<0.05).

The organizational structure of the forehead tissues has a strong effect on sound speed. Across a lateral plane, sound speeds are lowest in the center of the forehead. However, in the posterior region, the low sound speed center shifts toward the right (Fig. 6). This change in sound speed is evident within the melon tissue as well as between tissue types. From anterior to posterior, the sound speeds are higher toward the rostrum where the melon fat grades into blubber. This increase is also evident toward the front of the forehead in the dorsal–ventral direction. In general, a low sound speed core is present that converges posteriorly, exhibiting spatial asymmetry and strong sound speed gradients. Anteriorly, the core is broader in size/shape, spatially symmetric and exhibits a low gradient toward the dorsal rostral side. Laterally, the gradient is sharp throughout the melon.

Cuvier's beaked whale tissue densities ranged from 854 to 1175 kg m^–3. All tissues exhibited significantly different density values from one another except blubber and acoustic fats. Connective tissues were densest, followed by muscle and then the fatty tissues(Table 3; Fig. 5). Acoustic fats in the mandibular region were significantly lower than forehead acoustic fats and all blubber tissues (P≪0.001). Seawater with a salinity of 35‰,at 15°C and 101 kPa has a mean density of 1026 kg m^–3(Pilson, 1998). Fatty tissues(P<0.05) were significantly less dense than seawater, while muscle was not significantly different, and connective tissue was significantly denser than seawater (P<0.05).

Hounsfield units in Z. cavirostris tissues ranged from –106 to 108 HU, with sampling standard deviations ranging from 1 to 30 HU. All tissues exhibited significantly different HU values from each other(P≪0.001), except for acoustic fats and blubber. Again, connective tissues exhibited the highest values, followed by muscle and then fatty tissues (Table 3; Fig. 5). In the mandible,blubber HU values were significantly lower than those of acoustic fats and forehead blubber (P≪0.001). Blubber, acoustic fats and muscle values were significantly lower than those for water, while connective tissue exhibited HU values significantly higher than those of water(P<0.05).

Stress–strain curves were plotted for each of the samples from the data collected in the biomechanical compression tests(Fig. 7). Since the stress–strain curves were nonlinear, the elastic modulus (E)was evaluated at a variety of stresses to investigate how it changed with stress. These were averaged by tissue type and plotted against stress for the forehead and mandible tissues (Fig. 8). Muscle samples were discarded because myosin and actin connections, which are biologically important for retaining stiffness/structure, degrade with time after death(Schmidt-Nielsen, 1995).

Mean elastic modulus values for forehead and mandibular acoustic fat tissues ranged from 75 kPa to 934 kPa between stresses of 2.5 kPa and 50 kPa(Table 4). Blubber sample means from the forehead and mandibular regions ranged from 59 kPa to 1410 kPa at these stresses, and sample means for forehead connective tissue ranged from 124 to 1510 kPa. Across all stresses, acoustic fat samples had a lower elastic modulus for a given stress than did connective tissue and blubber. Blubber exhibits a nonlinear response in this strain range, becoming less compressible with increasing stress.

Percent hysteresis values in Z. cavirostris tissues ranged from 0.6 to 88.8%. Connective tissues exhibited significantly higher percent hysteresis than lipids (P≪0.001), while acoustic fats exhibited significantly higher percent hysteresis than blubber (P=0.01; Table 3). Lipids in the mandibular region had lower percent hysteresis than lipids in the forehead(P=0.03).

Principal components analysis was run to search for combinations of tissue properties that vary together systematically. The results show that the first principal component explained 72% of the variance, while the second component explained 23%. Sound speed, density and HU were explained by the first principal component and are therefore dependant measures(Fig. 9). Elasticity was an independent measure and is explained by the second component. The samples are distributed in a nonrandom pattern, with similar tissues clumping together as shown in Fig. 10. The plot shows that the connective tissues, muscle and lipids separate out along the first component. On the other hand, melon and blubber can be distinguished by the second component.

Since sound speed, density and HU are measuring a similar characteristic,we looked for the linear correlations between the measured values and determined r² values for HU versus density and sound speed. Both measures showed a strong positive relationship to HU(Fig. 11). The relationship between HU and sound speed was stronger (r²=0.85) than the one for HU and density (r²=0.76).

Several authors have discussed sound speed variations in the melons of marine Cetacea. Our measured sound speeds in the melon, ranging from 1412 to 1488 m s^–1 at room temperature(Table 3), are higher than those reported for other cetaceans, where most room temperature values are around 1370 m s^–1 (Apfel et al., 1985; Blomberg and Jensen, 1976; Blomberg and Lindholm, 1976; Litchfield et al., 1979; Norris and Harvey,1974). The study by Norris and Harvey(1974) had the most comparable values, with sound speeds ranging from 1273 to 1481 m s^–1. These differences may be due to technique: most studies extracted the oils from the tissues to perform sound speed tests, while our study and that of Norris and Harvey (1974) used the actual tissues. Another important difference of our study is that the subject was a neonate. Koopman et al. (2003a) and Lok and Folkersma(1979) have shown ontogenetic changes in beaked whale melon lipid composition, which may be reflected in changes of sound speed and transmission properties. However, sound speed differences may also reflect real differences between beaked whales and other cetaceans, as evidenced by differences in melon lipid chemistry(Litchfield et al., 1976).

Previous studies of sound speed in the odontocete forehead have reported a low sound speed inner core within the melon, consistent with the results of our study (Flewellen and Morris, 1987; Litchfield et al., 1979; Norris and Harvey, 1974). The presence of a low sound speed core supports the hypothesis that the beaked whale melon helps to produce a directional sound beam. The structure of the forehead, involving different tissues components, may give rise to stages of directional sound beam formation. For example, air spaces, dense connective tissue layers and bony regions can act as acoustically reflective surfaces that function to direct sound anteriorly. Another stage of beam formation results from the chemical topography in the melon, which tends to focus the beam along the low sound speed core.

The increase in sound speed between melon and blubber may be important in impedance matching of the sound waves to the water to maximize the acoustic energy transfer out of the forehead. Impedance matching occurs when the acoustic impedances (the product of density and sound speed) of two adjacent materials are similar, resulting in greater transmission of acoustic energy at the boundary between the two materials. Another factor that influences sound speed, and thereby impedance matching, is the change in temperature from the interior to exterior boundaries of the blubber. At the interior, sound speed averages 1401 m s^–1 for 37°C blubber, which increases to 1492 m s^–1 at the exterior blubber boundary in 15°C water(see Table 2), which has a sound speed of 1507 m s^–1. These results support acoustic impedance matching between the melon, blubber and seawater as a proposed function for the Cuvier's beaked whale melon, as proposed for the bottlenose dolphin melon (Norris and Harvey,1974). Sound speed in the fats in the mandibular channel also exhibit low sound speeds compared with surrounding tissues. Again, the change in sound speed from seawater to blubber to acoustic fat is probably important in impedance matching for sound reception and the channeling of sound from seawater to the ear complexes.

Sound speed values for cetacean blubber, muscle and connective tissue are not present in the literature; however, values for terrestrial mammal fats,muscle and collagen fibers (a main component of connective tissue) are available. Our Z. cavirostris sound speed values for blubber at 21°C (ranging from 1453 to 1489 m s^–1) are consistent with those reported for terrestrial mammals (1412–1489 m s^–1 for human, cow, dog and pig fat; reviewed by Duck, 1990). The sound speed values for most Z. cavirostris muscle samples at 21°C (ranging from 1461 to 1594 m s^–1) are also similar to those reported for terrestrial mammals (1542–1631 m s^–1 for human,cow, dog, pig and rabbit muscle; reviewed by Duck, 1990; O'Brien, 1977), but values for muscles found in the forehead region extend lower. Forehead region muscles appear to have increased lipid content compared with muscle tissues found throughout the body (Soldevilla et al,2004), which leads to a decrease in sound speed(Duck, 1990). Very little has been published on connective tissue sound speed; however, one study reports collagen sound speed at 1570 m s^–1 (reviewed in Duck, 1990; Lees and Rollins, 1972). Ziphius cavirostris connective tissue values exhibited a large range(1538–1708 m s^–1 at 21°C), which is probably due to the high variability in connective tissues that exhibit complex collagen structures. Collagen content in tissues affects sound speed significantly(O'Brien, 1977) such that the range of sound speed values may be associated with relative proportions of water and collagen in the samples.

The results of our temperature–sound speed experiments are similar to those reported by others. Our study of fats shows a linear decrease in sound speed with increasing temperature of ∼4 m s^–1deg.^–1, while muscle and connective tissue had a nearly constant slope. Duck (1990)reports that sound speed in fats decreases linearly with increasing temperature (at ∼10 m s^–1 deg.^–1) until∼35°C, where it exhibits nonlinear behavior and decreases at a lower rate. Studies have investigated sound speed changes with temperature in forehead acoustic fats in marine mammals. Lipids extracted from melon tissues decrease with increasing temperature, in the order of 1–13 m s^–1 deg.^–1(Blomberg and Jensen, 1976; Goold et al., 1996; Goold and Clarke, 2000; Litchfield et al., 1979). These studies (except Litchfield et al.,1979) have shown a nonlinear effect due to the phase change of the oils near body temperature. This is an important finding since beaked whale melon/acoustic fats have a different lipid composition, particularly in neonates, yet the lipids exhibit the same pattern of properties. We were not able to investigate nonlinear changes around body temperature since the tissues in our samples exhibited a phase change. It would be important to study this effect in beaked whale tissues in the future to better understand how the melon functions for sound conduction.

Mass densities of the Z. cavirostris tissue types are similar to those found in other mammals. Fatty tissues in this study ranged from 854 to 987 kg m^–3, averaging 922 kg m^–3. Duck(1990) reports mean human fat densities ranging from 917 to 939 kg m^–3 (human breast fat)and 990 to 1060 kg m^–3 (human breast including all tissues). The lower values that we found therefore correspond to pure fatty tissues,while the higher values include more structural tissues such as collagen in blubber. Our Z. cavirostris muscle values (ranging from 909 to 1066 kg m^–3) are similar to those reported for terrestrial mammals(ranging from 1038 to 1056 kg m^–3; Duck, 1990) but extend lower,suggesting the presence of lipids in forehead muscle tissues.

Ziphius cavirostris fats, including acoustic fats and blubber, had CT scan HU values ranging from –106 to –46. Robb(1998) found that mammalian fat values typically lie between –100 and –70. Most of our samples lie within this range; however, some blubber and melon fat samples from the forehead exhibited higher HU values than Robb found. This may be due to human error in matching the locations of the samples in the CT scan to their geometric space from dissection. It may also indicate that changes are due to the unique lipid chemistry of odontocete blubber and acoustic fats compared with common mammal fats or the presence of collagen in the blubber. This is a topic that merits further investigation.

Muscle HU values typically fall between 25 and 60 HU(Robb, 1998). Our values ranged from –61 to 2. These are lower than those reported by Robb, which may be caused by lipid content in the forehead muscle, which would decrease the HU value. Goodpaster et al.(2000) found that human skeletal muscle containing high lipid concentrations exhibited lower HU values than muscle that contained low lipid concentrations. The differences in our values might also be due to the amount of postmortem time and consequent decomposition in muscle tissues, perhaps the most susceptible to these changes.

Elastic modulus values for Z. cavirostris tissues range between 0.1 and 1.5 MPa at stresses from 2.5 to 50 kPa. Connective tissue and blubber exhibit the full range of values above, with elastic modulus increasing with increasing stress. Melon, on the other hand, exhibits lower values, with a high of 0.93 MPa at 50 kPa. The higher values for blubber and connective tissue suggest that these tissues have greater structure than melon, which is reasonable since connective tissue and blubber are held together by collagen fibers. O'Brien (1977) notes that the elastic properties of soft tissue are determined primarily by the content of collagen and other structural proteins. Elastic moduli for tissues vary nonlinearly with applied stress, making comparisons with previously published values difficult as these measurements may have been carried out on different portions of the stress–strain curve. Percent hysteresis values for Z. cavirostris fall in the typical range for mammalian tissues. Collagen exhibits low values (7%), while materials like silk exhibit high values (65%), with the majority of tissues falling around 20% (Vogel,1988).

The results of the PCA show that connective tissue, muscle and lipids separate distinctly along the first principal component(Fig. 10), which represents sound speed, density and HU (Fig. 9). This separation shows that sound speed, density or HU can each be used to determine the type of tissue in a Cuvier's beaked whale, since each tissue type's physical properties are distinct. The high explanatory power of the first principal component illustrates that sound speed, density and HU are measuring similar characteristics. This supports the notion that HUs can be used to model density and sound speed parameters needed for acoustic modeling,which is beneficial to modeling efforts, since HUs can be collected throughout a whole specimen non-destructively. Hounsfield units cannot be used to predict elastic modulus values. Elasticity, primarily represented by the second principal component, is the only measure that distinguishes between acoustic fats and blubber (Fig. 10). Elasticity is important for understanding the behavior of these tissues under high stress conditions.

The linear correlations between HUs and sound speed and density were high(0.85 and 0.76, respectively). The linear relationships we found for these measures provide a model to determine the sound speed and density when only the HUs from CT scanning are available(Fig. 11). Some of the noise in the data may be due to human error in correlating the samples to their location in the CT scans. Future studies should work to improve this method.

Our analysis of the physical properties of Cuvier's beaked whale tissues provides improvements on past research and insight into sound generation,reception and acoustic trauma modeling. However, the results do have some limitations. (1) Sound speed measurements were made at 10 MHz, while the frequencies of interest range from 1 to 200 kHz. Sound speed dispersion for biological soft tissues and marine mammal oils ranges from 1 to 10 m s^–1 between 1 and 10 MHz (Cartensen and Schwan, 1959; Kremkau et al., 1981; Kuo and Weng, 1975; O'Donnell et al., 1981). While our values may be a few m s^–1 too high, this is within the error of our measurements. (2) The specimen we studied was a neonate. H. N. Koopman, S. M. Budge, D. R. Ketten and S. J. Iverson (personal communcation)have described ontogenetic changes in the chemical makeup of acoustic fats in beaked whales that may affect physical properties of tissues. This is an important distinction for sound generation and reception modeling. However, Ziphius cavirostris specimens are not readily available. The present study is the first to present any physical property data from a Cuvier's beaked whale. (3) Temperature–sound speed measurements were not carried out at 37°C, our estimated in vivo temperature. Sound speed in lipids exhibits nonlinear changes around body temperature, suggesting that our calculated sound speeds may be a few m s^–1 too low. It would be important to study this effect in beaked whale tissues in the future to better understand how the melon functions for sound conduction. (4)Histological analysis of tissue categories was beyond the scope of our study. Variations in tissue content of each sample could have led to the variation we see in our results. (5) Measurements were made on thawed, postmortemtissues. Changes in tissue physical properties may arise after death, and with freezing. No significant changes were found for melon fat, blubber and connective tissue CT scan HU values in live, recently deceased and frozen bottlenose dolphin specimens (M. F. McKenna, personal communication). No significant differences have been found for sound speeds in bottlenose dolphin tissues as a function of time after death and freezing (M. F. McKenna,personal communication). The same results were found for sound speeds in fresh and frozen mammalian liver (Van der Steen et al, 1991), myocardial tissues(Dent et al., 2000) and human breast tissue (Foster et al.,1984). Elastic properties of thawed non-contractile tissues,including human intervertebral discs and rabbit ligaments, are not significantly different from fresh samples(Smeathers and Joanes, 1988; Woo et al., 1986). Fitzgerald(1975) and Fitzgerald and Fitzgerald (1995) suggest there is a life-to-death transition in visco-elastic compliance for beef fat,porcine intervertebral disc and whale blubber around 5–11 h postmortem. Significant changes in the elastic properties of contractile muscle tissues result from the postmortem and freezing process (Leitschuh et al.,1996; Gottsauner-Wolf et al.,1995; Van Ee et al.,2000). In summary, sound speed and CT scan HU values are not significantly affected by postmortem and freezing processes. Elastic properties of non-contractile tissues may exhibit postmortemdifferences but are not affected by freezing, and elastic properties of contractile tissues are significantly affected by postmortem and freezing effects. For this reason, we have discarded muscle from our analysis of elastic properties.

This study provides high topographical resolution data on sound speed, mass density, CT scan Hounsfield units and elasticity for Z. cavirostristissues. This has been the first attempt at gathering this breadth of physical property data from beaked whale tissues. We have examined the similarities and differences that exist between these properties from Cuvier's beaked whale tissues, which have unique structural and chemical characteristics, and those found in other marine and terrestrial mammals. Evidence for a relationship between CT scan HUs and sound speed and mass density is presented. This may allow future studies to relate CT scan values from Z. cavirostris to these physical properties without extracting tissue samples. While there are limitations to our methods and results, the data we present show promise for use in acoustic-structural models of a beaked whale that may provide insight into sound emission and reception paths and into possible mechanisms for Z. cavirostris' sensitivity to sound.

We thank: Susan Chivers, NMFS, for providing access to the beaked whale specimen; Hubbs Sea-World for allowing us use of their dissection lab; Lisa Cooper and Liliana Fajardo for aiding in the dissection; Shultz Steel for allowing use of their band-saw to section the specimen; Mark Shaver at UCSD Medical Center for help with CT scanning; Dr Thomas Nelson for obtaining CT data and providing eFilm assistance; Dr Richard Buxton for allowing use of his MR scanner; Cecelia Kemper for help with the MR scanner; Graydon Armsworthy,Jeanine Donley, Chris Garsha, Jeremy Goldbogen, Scott Rappaport, Allan Sauter and Judy St Ledger for all of their assistance throughout the study. We thank two anonymous reviewers whose comments greatly improved this manuscript. This work was supported by the US Navy CNO-N45, with project management by Frank Stone and Ernie Young.