‘Macrosmatic’ mammals have dedicated olfactory regions within their nasal cavity and segregated airstreams for olfaction and respiratory air-conditioning. Here, we examined the 3D distribution of olfactory surface area (SA) and nasal airflow patterns in the pygmy slow loris (Nycticebus pygmaeus), a primate with primitive nasal cavities, except for enlarged eyes that converge upon the posterodorsal nasal region. Using the head of an adult loris cadaver, we co-registered micro-computed tomography (CT) slices and histology sections to create a 3D reconstruction of the olfactory mucosa distribution. Histological sections were used to measure olfactory surface area and to annotate CT reconstructions. The loris has a complex olfactory recess (∼19% of total nasal SA) with multiple olfactory turbinals. However, the first ethmoturbinal has a rostral projection that extends far anterior to the olfactory recess, lined by ∼90% non-olfactory epithelium. Only one (of three) frontoturbinals bears olfactory mucosa. Computational fluid dynamics simulations of nasal airflow and odorant deposition revealed that there is some segregation of respiratory and olfactory flow in the loris nose, but that it is not as distinct as in well-studied ‘macrosmats' (e.g. the dog). In the loris, airflow is segregated medially and laterally to vertically elongated, plate-like first ethmoturbinals. Thus, lorises may be said to have certain macrosmatic anatomical characteristics (e.g. olfactory recess), but not segregated nasal airflow patterns that are optimized for olfaction, as in canids. These results imply that a binary ‘microsmatic/macrosmatic’ dichotomy does not exist. Rather, mammals appear to exhibit complex trends with respect to specialization of the turbinals and recesses.
To a varying degree, all primates employ olfaction in social interactions or in association with feeding functions (Nekaris, 2005). However, genetic evidence indicates that monkeys, apes and humans have undergone a functional reduction in certain olfactory genes (e.g. Rouquier et al., 2000). Moreover, in humans, the airflow patterns differ greatly from those of mammals with a more primitive nasal cavity. In particular, humans and other haplorhines (collectively, anthropoids and tarsiers) lack an olfactory recess, a posterior cul-de-sac of the nasal fossa that is present in most terrestrial mammals (Van Valkenburgh et al., 2014; Smith et al., 2015). In mammals possessing this space, odorants migrate through the olfactory region more slowly and are thus more likely to be detected (Yang et al., 2007; Craven et al., 2010; Eiting et al., 2014).
To date, nasal airflow has been studied in few primates. In this study, we examined a nocturnal primate, the pygmy slow loris (Nycticebus pygmaeus), using computational fluid dynamics (CFD) simulations of airflow and odorant deposition. Nycticebus, like other lorises and lemurs, has a primitive nasal cavity with four medial turbinals that are at least partially covered with olfactory mucosa. Like other lorises and lemurs, a portion of the ethmoturbinal complex is contained within the olfactory recess (DeLeon and Smith, 2014). Chemical communication via urine and glandular secretions is important in territorial and other social interactions for these arboreal primates (Fisher, 2003; Hagey et al., 2007).
All living strepsirrhine primates (lemurs and lorises) have a more complicated olfactory anatomy than living anthropoids (monkeys, apes and humans). However, the entire order Primates has evolved as visual specialists (Barton, 1998). Living primates possess one or more correlates of visual specialization that may physically encroach on the interorbital (i.e. nasal) space during development. First, in many primates the eyes are relatively large compared with most other mammals (Ross and Kirk, 2007). Second, the eyes are convergent toward the midline in all primates, to a varying degree, for binocular vision (Heesy, 2009). Because eyes do not scale isometrically with the rest of the body (Ross and Kirk, 2007), these ocular traits create a bigger ‘packaging’ problem in small-bodied primates. In general, the eyes encroach on the nasal region to a greater degree than in large-bodied species, but this is less significant in strepsirrhines because the eyes are partially ectopic and may be positioned lateral or dorsolateral to the nasal fossa (Cartmill, 1972).
Nycticebus is a strepsirrhine with exceptionally large eyes and small body size. In the present study, we sought to determine the degree to which airflow in the nasal fossa is optimized for odorant delivery to the olfactory recess. In particular, we investigated the degree to which nasal airflow patterns resemble those seen in the dog and other macrosmatic animals. In the dog, olfactory airflow is shunted dorsally to the olfactory recess, rapidly bypassing more ventral structures such as the complex maxilloturbinal; respiratory streams are segregated ventrally (Craven et al., 2010; Lawson et al., 2012). We also considered the significance of relatively large eyes in lorises. These are known to influence bone resorption and medial drift at shared boundaries of the orbital and nasal cavities during development (Smith et al., 2014), and thus affect nasal fossa configuration and possibly respiratory and olfactory function.
MATERIALS AND METHODS
Specimen information and imaging methods
An 11 year old male Nycticebus pygmaeus Bonhote 1907 that died in captivity at the Cleveland Metroparks Zoo was studied. After necropsy and removal of the brain for an unrelated study, the remainder of the head was immersed in formalin. Use of animal tissues was reviewed and approved by the IACUC at Slippery Rock University. The head was scanned at Northeast Ohio Medical University (NEOMED) using the Scanco vivaCT micro-computed tomography (CT) scanner (parameters: 25 µm cubic voxels; 70 kVp, 114 µA). Subsequently, the head was processed for paraffin histology. Serial sections at 12 µm thickness were made and alternating sections were stained with Gomori Trichrome or hematoxylin–eosin procedures.
The CT image stack was then imported into Amira software to generate a 3D reconstruction of the bony nasal fossa. The CT data of the skull were reconstructed and resliced to match the cross-sectional plane of histology. The full details of this method are described elsewhere (DeLeon and Smith, 2014). Briefly, we observed corresponding anatomical features in the CT and histological data, and used these to adjust the plane of section in the CT data. The CT dataset was then digitally re-sliced, so that the plane of CT slices corresponded to the histological sections.
Next, we segmented the lumen of the nasal airway based on histological sections of the right nasal fossa. Micrographs of every tenth histological section were segmented using Adobe Photoshop and saved as a stack of bitmap images. In cases where a histological section was damaged, an adjacent stained section was prepared from archived paraffin sections. The series of segmented images were auto-aligned using ImageJ with manual alignment used as necessary. Achieving congruity of cross-sectional planes in CT and histology allowed us to correct artefactual distortions that are attributable to the decalcification and ethanol baths used for paraffin histology. In particular, the free margins of the ethmoturbinals, which project far rostral from their basal lamellae, undergo shrinkage (DeLeon and Smith, 2014). To transform the histology images to correspond to the CT images, each segmented histology image from the ethmoturbinal region was registered to the corresponding CT slice using Adobe Photoshop. Whenever a turbinal structure was observed in the CT slice but not in the corresponding segmented histology image, the image file of a more posterior histology section containing that turbinal structure was also imported, and the turbinal was cut and pasted into alignment corresponding to the CT image. This process of expanding the rostrocaudal dimension of the turbinal, as visualized through histology, to correspond to the CT images resulted in CT slices with no corresponding histology for the turbinal. Additional intervening histological sections were mounted, stained and photographed to provide the corresponding histology for those CT slices. This avoided excessive interpolation of interslice distance as the turbinal length was corrected.
We used the histological sections to determine olfactory mucosa distribution. This was done by microscopic observations on sections with a Leica DMLB photomicroscope at ×50 to ×200 magnification. Digital micrograph images of each section were simultaneously projected on a second screen to annotate according to epithelium type. First, the levels where only squamous epithelium exists (e.g. in the vestibule) were indicated anteriorly, because this epithelium is comparatively impermeable to odorant. Second, the olfactory epithelium in the nasal fossa was ‘mapped’ by annotating olfactory portions of the ethmoturbinals in each section where olfactory mucosa is present. Olfactory epithelium was identified based on the presence of olfactory receptor neurons, which usually occur in multiple rows and whose nuclei are larger and rounder than those of adjacent respiratory epithelium (see Smith et al., 2007, for further information).
With this information, we annotated the segmented bitmap images that were used in the airflow simulations according to olfactory/respiratory mucosa boundaries on ethmoturbinals. In addition, we created mucosal ‘masks’ to import into Amira. Briefly, this involved aligning binary images of ethmoturbinal contours with CT slices using Adobe Photoshop. This was done separately for total mucosal contours and then for olfactory mucosa only. Using Amira software, we reconstructed the bony anatomy of the internal nasal fossa and then superimposed all mucosa and olfactory mucosa. The final results of these procedures were two models: (1) a segmented image stack of the right nasal airway lumen with the contours delineated by epithelium type (squamous, olfactory or respiratory), and (2) a 3D surface model of the skeletal anatomy of the nasal cavity, color coded to denote the olfactory mucosa distribution on the ethmoturbinal complex.
Finally, surface areas of mucosa type, including olfactory and non-olfactory, were quantified using ImageJ for a comparison to previous reports on primates. The annotated images of histology sections were used. After setting the scale based on a micrograph of a stage micrometer at the same magnification as the sections, two measurements were taken using the freehand tracing tool: the total perimeter of the nasal mucosa and the perimeter of the olfactory mucosa. These perimeter measurements (in mm) were multiplied by the distance between adjacent sections to compute an approximate segmental surface area. All segmental areas were summed to compute the total surface area of the right nasal fossa. The surface areas of individual turbinals and spaces were also calculated (more details are provided in Smith et al., 2011). All 3D reconstructions and quantifications are based on corrected rostrocaudal dimensions of all ethmoturbinals as described above.
Given the segmented histological sections that were annotated by epithelium type (squamous, olfactory and other), a 3D surface model of the right nasal airway was reconstructed as in previous work (Craven et al., 2007; Ranslow et al., 2014; Coppola et al., 2014; Pang et al., 2016). The anatomical reconstruction extends from the first histology section, located in the nasal vestibule approximately 2.8 mm from the tip of the nose, to the nasopharynx (Fig. 1). Importantly, the reconstruction includes a specific delineation of the nasal epithelium, allowing us to investigate airflow and odorant deposition in sensory and non-sensory regions of the nose.
Given the reconstructed surface model, two unstructured hexahedral computational meshes (coarse and fine) were generated using the snappyHexMesh utility available in the open-source computational continuum mechanics library OpenFOAM (version 2.4). The coarse and fine meshes (Fig. 1C,D) contain approximately 5.6 million and 11.2 million computational cells, respectively, including five wall-normal layers to resolve the large velocity and odorant concentration gradients that occur at the wall.
CFD simulations of quasi-steady inspiratory airflow and odorant deposition were performed using OpenFOAM as described by Rygg et al. (2017). Briefly, the SIMPLE (Semi-Implicit Method for Pressure-Linked Equations) algorithm was used to numerically solve the governing equations for the steady-state incompressible flow of air in the nose using second-order accurate spatial discretization schemes. A uniform velocity inlet boundary condition was specified at the naris to obtain physiologically realistic flow rates for two conditions: quiet breathing and a quasi-steady sniff. The airflow rate for breathing through a single nostril (7.9 ml s−1) was approximated based on the average adult body mass for the species (0.42 kg; Rowe and Myers, 2013) and the allometric equation for mean inspiratory flow rate provided by Frappell et al. (1992). The airflow rate through a single nostril during a quasi-steady sniff was estimated by comparing respiratory and sniffing flow rate measurements for an animal that is of similar size to the loris, the rat. Specifically, we compared the average inspiratory flow rate measurements of Frappell et al. (1992) for breathing in the rat with the measurements of Youngentob et al. (1987) for the peak flow rate during sniffing in the rat, yielding a multiplicative factor of 1.6 between the two physiological conditions (breathing and sniffing). That is, based on available experimental data, the maximum flow rate during sniffing in the rat is about 1.6 times greater than the average flow rate for respiration. We then utilized this multiplicative factor of 1.6 to calculate an estimate for the sniffing flow rate in the loris of 12.7 ml s−1. To further explore the sensitivity of our results to these flow rate estimates, we also simulated two other flow rates: 4.1 and 8.4 ml s−1.
Converged steady-state airflow simulations were obtained by ensuring that the normalized solution residuals were less than 1×10−6 and by monitoring the iterative convergence of various computed quantities (e.g. volumetric flow rate, maximum velocity, minimum and maximum pressure). Given the computed steady-state airflow solutions for breathing and sniffing, simulations of odorant deposition were performed for three odorants: heptanoic acid, isoamyl acetate and nonane. These three specific odorants were chosen because they are highly soluble, moderately soluble and insoluble, respectively, in the mucus layer that lines the nasal cavity. The required odorant properties were obtained as described by Rygg et al. (2017) and CFD simulations of quasi-steady odorant deposition were performed as in previous work (Rygg et al., 2017; Coppola et al., 2017, 2019).
To investigate the numerical accuracy of the computed solutions, we performed a mesh refinement study at the sniffing flow rate. Applying the same inlet flow rate boundary condition, the overall pressure drop between the naris and nasopharynx for the coarse and fine meshes differed by only 1.4%, indicating that the CFD solution is fairly insensitive to mesh resolution at this level of refinement. For all of the simulations reported in this study, we utilized the fine mesh that contains approximately 11.2 million computational cells.
The mucosal maps revealed that most olfactory mucosa on the ethmoturbinals is restricted, approximately, to the posterior third of this entire complex of turbinals. Almost all of the olfactory mucosa on the ethmoturbinals is restricted to the olfactory recess, defined as the space dorsal to the transverse lamina (Fig. 2). The first ethmoturbinal is primarily covered with non-olfactory epithelium (∼90%; Table 1). It has a free projection that extends rostrally to overlap the maxilloturbinal; none of this part of ethmoturbinal I bears olfactory mucosa. As previously observed (Le Gros Clark, 1959; Smith et al., 2007), this turbinal also extends deeply inferiorly, near the palate. Our reconstruction of N. pygmaeus shows the medial lamina of the first ethmoturbinal descends near the floor of the nasal fossa, and parallel to the posteroinferior-most limit of the maxilloturbinal. Both structures are visible when looking through the nasopharyngeal ducts from the posterior side (Fig. 2).
The olfactory recess is complex in that all four ethmoturbinals are located at least partially within it. In addition, one interturbinal begins anterior to the olfactory recess and then tucks within it, between ethmoturbinals II and III. In sum, about 19% of the total nasal surface area is within the olfactory recess. Moving posteriorly, the ethmoturbinals bear progressively more olfactory mucosa (Fig. 2). Ethmoturbinal IV is lined with 24.2% olfactory mucosa. The frontal recess has three frontoturbinals within it, but only the inferior-most of them bears any olfactory mucosa (46%; Table 1).
CFD simulations of airflow revealed similar gross inspiratory nasal flow patterns for both respiration and sniffing. In both cases, the highest flow speeds in the nasal cavity are primarily along the septum in the common meatus (Fig. 3A). Generally, the flow within the dorsal meatus is directed to the olfactory recess. However, some flow streams outside of the dorsal meatus also reach the olfactory recess (Fig. 3B). Thus, while there is some segregation of respiratory and olfactory flow in the loris nose, it is not as distinct as in the dog (e.g. Craven et al., 2010).
CFD simulations of odorant deposition showed that there is not a large difference in the deposition patterns for respiration and sniffing flow conditions, particularly within the olfactory recess (Fig. 4). The highly soluble odorant heptanoic acid is deposited in the respiratory region and in the anterior olfactory recess (Fig. 4A,B). The insoluble odorant nonane is deposited more uniformly in both regions (Fig. 4E,F). There is a larger gradient of odorant flux in the nose for the moderately soluble isoamyl acetate (Fig. 4C,D). However, within the olfactory region, most of the isoamyl acetate vapor is deposited within the anterior half of the recess, along the septum and on the medial side of the ethmoturbinals. Comparing the gross deposition patterns for heptanoic acid and isoamyl acetate in the olfactory recess (Fig. 4A,B versus C,D), there is a modest difference, but it is not as substantial as the marked difference in deposition patterns observed between highly and moderately soluble chemicals in the dog (Lawson et al., 2012) and more recently in the mouse (Coppola et al., 2017).
Our flow rate sensitivity study confirmed that the present nasal airflow and odorant deposition patterns are insensitive to our estimates of the respiratory and sniffing flow rates for the loris. Specifically, the additional CFD simulations performed at 4.1 and 8.4 ml s−1 revealed extremely similar nasal airflow and odorant deposition patterns to those observed for respiration and sniffing. Thus, the present results are expected to be generally representative for the loris across a wide range of nasal flow rates, from quiet breathing to the higher flow rates that occur during sniffing.
Much of our detailed knowledge of primate olfactory anatomy is rooted in studies on humans or other anthropoid primates, all of which possess relatively less-complex ethmoturbinal morphology, and less surface area of olfactory neuroepithelium compared with most other mammals (Harkema et al., 2006). These traits have relegated primates such as ourselves to so-called microsmatic animals, or those with a comparatively poor sense of smell (Turner, 1891; Negus, 1958; but see Laska et al., 2000). Strepsirrhine primates have not been as closely studied. Although their internal nasal architecture is more complex and more extensively covered with olfactory mucosa compared with that of anthropoids (Smith et al., 2011, 2016), a broader comparative understanding is lacking. Because their nasal anatomy more closely resembles that of early primates, a detailed analysis of their nasal physiology may provide clues for the evolution of the primate nose and face.
Rodents and dogs are among the most commonly studied mammals regarding nasal airflow and olfaction. In both, the nasal cavity includes a dorsoposterior cul-de-sac, termed the olfactory recess. This recess communicates anteriorly with the main nasal chamber. Computer simulations reveal that air enters this recess via the dorsal meatus and then slowly filters through the convoluted airways between ethmoturbinals (Yang et al., 2007; Craven et al., 2010; Coppola et al., 2017). Computer modeling has also been used to study the influence of the functional size of the olfactory recess by virtually altering the extent of the transverse lamina, which showed that a larger functional recess is potentially advantageous for olfactory odorant transport (Eiting et al., 2014). In most mammalian orders, at least some species possess this specialized chamber. In Primates, all extant strepsirrhines and perhaps some haplorrhine ancestors possess it (Smith and Rossie, 2006; Kirk et al., 2014).
In the mouse lemur (Microcebus murinus), this space is demonstrably dedicated to olfactory function in that ∼69% of the olfactory recess is lined with olfactory mucosa, whereas only ∼28% of the remainder of the nasal cavity (excluding the recess) is lined with olfactory mucosa (Smith et al., 2011). Similarly, our 3D reconstruction shows that the majority of the olfactory mucosa on ethmoturbinals is contained within the olfactory recess in N. pygmaeus (Fig. 1). In this way, they are both similar to rodents and dogs. However, a separate question is whether the strepsirrhine nasal apparatus has segregated respiratory and olfactory airflow streams, as in the dog (Craven et al., 2010). The present CFD simulations reveal that there is some segregation of respiratory and olfactory flow in the loris nose (Fig. 3B), but that it is not as distinct as in the dog (Craven et al., 2010; Rygg et al., 2017) and mouse (Coppola et al., 2017). Moreover, the simulations show that, unlike the dog, the dorsal meatus in the loris does not contain high-speed airflow (Fig. 3A). In dogs, the dorsal meatus is the only route by which air enters the olfactory recess and high flow speeds minimize odorant deposition in the upstream respiratory region (Craven et al., 2009; Rygg et al., 2017). In contrast, the lack of high-speed flow in the dorsal meatus of the loris may explain why there is not a substantial difference in odorant deposition patterns between respiration and sniffing, or between the highly soluble heptanoic acid and the moderately soluble isoamyl acetate (Fig. 4). We speculate that this might influence olfactory function by reducing the amount of ‘imposed’ patterning in the loris nose for different odors compared with the dog, which is thought to be used in concert with the ‘inherent’ patterning of olfactory receptors for odor recognition (Lawson et al., 2012).
These findings suggest that strepsirrhine primates do not have the degree of macrosmatic specialization seen in dogs, in which internal nasal anatomy is organized to deliver a functionally distinct airstream optimized for odorant delivery to the olfactory recess. Dedicated respiratory and olfactory flow streams in the dog's nose correspond to segregated nasal anatomy: respiratory and olfactory turbinals are in relatively non-overlapping regions (Craven et al., 2010; Pang et al., 2016). Thus, a distinction between the most extreme olfactory specialists and other mammals may be recognizable in the extent to which ethmoturbinals are spatially segregated from the main chamber, and from the maxilloturbinal. In this sense, all strepsirrhines studied to date diverge from the extreme macrosmatic form in important ways. First, there is far more overlap of maxilloturbinals and ethmoturbinals in strepsirrhine primates (Smith et al., 2007; 2016) than has been observed in rodents and dogs (Adams, 1972; Craven et al., 2007). Second, ethmoturbinal I is enlarged and largely overlapping the maxilloturbinal. The overlap of ethmoturbinal I and the maxilloturbinal is also characteristic of scandentians and dermopterans, which, with primates, comprise the living euarchontans (Smith et al., 2015). This suggests that the arrangement may be primitive for primates.
The majority of ethmoturbinal I in all primates is known to be non-olfactory (Smith et al., 2007, 2011, 2016), indicating its most important functions may be humidifying inspired air or directing airstreams. In lorisoids and some other strepsirrhines, ethmoturbinal I has a large medial plate which is vertically oriented (Smith et al., 2007). This may explain medial versus lateral differences in airflow and the associated disruption of respiratory and olfactory flow segregation observed here. The air medially adjacent to the septum is likely segregated by the vertical plate of ethmoturbinal I (Fig. 3B), which is large and non-olfactory in N. pygmaeus (Fig. 2, Table 1).
The paranasal spaces are also distinctive in the loris compared with those described in other strepsirrhines. In particular, we suggest that the frontal recess may be diminished developmentally by the encroaching orbits (see Fig. 2). There is empirical developmental evidence for how a trade-off may occur between the shape of the orbital and nasal cavities. Bone modeling patterns were recently examined in a fetal slender loris (Loris tardigradus); patterns of osteoclast activity revealed medial resorption within the developing fetal orbit (Smith et al., 2014). Specifically, bone cell activity indicates that the frontal bone, a shared skeletal boundary of the orbit and frontal recess, drifts medially during growth. This suggests that the enlarged eyes of this nocturnal species may constrain the size of the paranasal spaces during development. Similar constraints may explain a relatively reduced frontal recess in N. pygmaeus because the orbit is greatly enlarged at its interface with the dorsolateral nasal cavity (Fig. 2).
Compared with the remainder of the nasal fossa, olfactory function of the paranasal spaces is less well understood. Yet, many mammals possess olfactory turbinals within this space. The results of the present study concur with a recent study by Rygg et al. (2017) on the coyote (Canis latrans). In N. pygameus and C. latrans, the solubility of odorants determines where deposition occurs. The least soluble among the tested odorants (nonane) was the only one to reach the frontal recess in both species. This suggests that the frontal paranasal space may only be sensitive to some odorant types that are relatively insoluble. Despite the similarity, N. pygmaeus possesses relatively little olfactory mucosa in the paranasal space compared with the olfactory recess; only one of the three frontoturbinals bears olfactory mucosa. An interesting implication is that N. pygmaeus and perhaps some other strepsirrhines may have a reduction in olfactory surface area for detecting some odorants as a consequence of orbital growth patterns.
Broader comparative evidence may be needed to fully understand whether olfactory specialization reflects a process of segregating the maxilloturbinal and ethmoturbinals from one another. Whereas strepsirrhine primates have more complex turbinals than humans, the so-called ‘olfactory’ turbinals (ethmoturbinals) are not specialized for olfaction to the degree seen in dogs or rodents. Furthermore, comparison of N. pygmaeus and M. murinus shows that there may be considerable variability in the degree to which ethmoturbinals participate in non-olfactory functions. For example, about 2/3 of the surface area of ethmoturbinal I is non-olfactory in M. murinus, compared with 90% in N. pygmaeus (Table 1). Our observations also reveal a contrast in the olfactory recess in these species. Much less of the ethmoturbinal complex is contained in the recess in M. murinus (ethmoturbinal IV only) compared with N. pygmaeus (parts of all ethmoturbinals). Microcebus murinus likely relies more heavily on airflow in the main chamber for odorant detection compared with N. pygmaeus; indeed, the olfactory mucosa projects far beyond the reaches of the recess in the former. These variations suggest that the presence of an olfactory recess alone does not imply complete segregation of respiratory and olfactory flow streams in the nose.
The primitive arrangement for mammals is not completely clear as yet. The olfactory recess is an early innovation in mammals, present in at least some mammalian cynodonts (Ruf et al., 2014). However, the delicate nature of the turbinal lamellae, as well as their initially cartilaginous composition, make it uncertain how far rostrally the ethmoturbinals projected from within the recess (Crompton et al., 2017). Spatial overlap of the maxilloturbinal and the ethmoturbinal complex is minimal in some groups of mammals, including canids (Craven et al., 2007), many ungulates (Negus, 1958; Ranslow et al., 2014), and rodents (e.g. Adams, 1972). However, the ethmoturbinals (particularly the first) project far rostrally from the olfactory recess in a diverse array of mammals, including known felids (Pang et al., 2016), at least some marsupials (Rowe et al., 2005), as well as many living euarchontans (Smith et al., 2015) and at least some fossil primates (Lundeen and Kirk, 2019). The widespread nature of this arrangement across mammals is suggestive of a primitive condition. If so, then those mammals in which the ethmoturbinals are more limited to the olfactory recess, with less projection of ethmoturbinal I, possess a derived morphology. The macrosmatic segregation of respiratory and olfactory regions might have occurred convergently in living rodents, ungulates and canids via rearrangement and spatial segregation of turbinals.
Finally, we note a limitation of this study and prospects for future work. Because the present CFD model was reconstructed from segmented histology sections, the model did not include the external nose. The neglect of the external nose and the use of a uniform velocity boundary condition at the truncated naris located approximately 2.8 mm from the tip of the nose may have slightly influenced the downstream flow field, but is unlikely to have influenced the gross nasal airflow and odorant deposition patterns reported here. Even so, future work should leverage high-resolution magnetic resonance imaging (MRI) or CT data to reconstruct the tissue-lined nasal cavity that includes the external nose (e.g. Craven et al., 2009; Ranslow et al., 2014). Diffusible iodine-based contrast-enhanced CT (diceCT) (Gignac et al., 2016) might be used to resolve the nasal epithelium, but further work is needed to verify that it can be used to reliably delineate sensory from non-sensory nasal epithelium. Alternatively, tissue thickness has been used by Yee et al. (2016) to delineate sensory and non-sensory epithelium, which may also be explored as a means to map the nasal epithelium from high-resolution CT data alone. This approach, however, will need to be validated by comparison with histology data in the same specimen.
The findings on nasal morphometry and airflow patterns in N. pygmaeus indicate that this strepsirrhine primate does not have the degree of macrosmatic specialization seen in canids, in which internal nasal anatomy is organized to deliver a functionally distinct airstream optimized for odorant delivery to the olfactory recess (Craven et al., 2010; Rygg et al., 2017). In N. pygmaeus, this appears to be due to the large vertical lamina of ethmoturbinal I that partitions airflow vertically into medial and lateral streams. Whether the large degree of overlap of the ethmoturbinal complex with the maxilloturbinals is primitive for mammals remains a critical unanswered question. In addition, primate trends in evolution of the frontal recess, which may detect a specific range of odorants, require elucidation.
Finally, our results imply that in N. pygmaeus and other strepsirrhines, the ethmoturbinal complex is multifunctional to a degree not seen in canids. While in canids, nasal morphology may be said to reflect an optimized design for distinct respiratory and olfactory airflow streams (Craven et al., 2010), in strepsirrhines the ethmoturbinals presumably play more significant roles in directing airflow and in humidifying and warming inspired air (the latter in combination with the maxilloturbinal). These results imply that a binary microsmatic/macrosmatic dichotomy does not exist. Rather, mammals appear to exhibit a diversity in nasal morphology that lies along a continuum characterized by the degree to which respiratory and olfactory anatomy and function are segregated in the nose. Future work should consider the evolutionary origins of this diversity.
The authors thank C. J. Vinyard for scanning the cadaveric head used in the present study.
Conceptualization: T.D.S., B.A.C., V.B.D.; Methodology: T.D.S., B.A.C., C.J.B., V.B.D.; Formal analysis: T.D.S., B.A.C., S.M.E., V.B.D.; Investigation: T.D.S., B.A.C.; Resources: C.J.B.; Writing - original draft: T.D.S., B.A.C.; Writing - review & editing: T.D.S., B.A.C., S.M.E., V.B.D.; Funding acquisition: T.D.S., B.A.C., V.B.D.
This study was supported by the US National Science Foundation [grant numbers: BCS-1231717, BCS-1231350, IOS-1120375].
The authors declare no competing or financial interests.