Terminology

We note that the terms concha and turbinate have been used interchangeably for these structures in birds (e.g., King 1993 vs. Geist 2000). When applied to mammals, turbinate refers to the bony structure, whereas concha refers to the turbinate and tissue covering it. In this paper, we use the term concha to refer to the prominent, bilaterally paired cartilaginous structures that project into the nasal cavities of birds. We describe anatomical planes based on a bird standing in a natural posture and holding its bill roughly parallel with the ground.

Study Taxa and Specimens

To study the nasal conchae with CT scans, we used fluid-preserved specimens, including 8 M. m. atlantica (2 females, 6 males) collected on the Delmarva Peninsula, Delaware, and 7 M. m. melodia (1 female, 6 males) collected near Washington, DC, during the breeding season (July) of 2013. Following collection, all specimens were frozen in air-tight bags and later fixed in formalin and then transferred to 70% ethanol. Before imaging, specimens were transferred to 100% ethanol for 2 nights to allow adjustment to the new ethanol concentration. They were then soaked in a solution of 1 part iodine to 99 parts of 100% ethanol for 32 or 44 days so that soft tissues and bones could be simultaneously visualized in CT scans. Because specimens shrink when preserved in ethanol and iodine (Vickerton et al. 2013), we took precautions to control for and minimize this shrinkage. First, we incubated specimens in 70% ethanol for >14 days, after which tissue size stabilizes (Vickerton et al. 2013), as recommended by Gignac et al. (2016). Second, we used a minimal concentration of iodine. Third, we did not directly compare specimens preserved in ethanol and iodine to those dried and stuffed. We measured the tarsus (tarsometatarsus) with calipers (precision 0.1 mm) and wing with a wing rule (precision 1 mm). We determined sex by breeding condition and wing length (Pyle 1997).

For additional measurements of nasal cavity dimensions, we radiographed dry stuffed bird skins, consisting of 20 M. m. atlantica collected on the Delmarva Peninsula and 19 M. m. melodia collected near Washington, DC, all of which are housed at the Smithsonian's National Museum of Natural History. All skins were prepared from birds collected during the breeding season (April–July) between 1876 and 2001 and were identified by the collector as adult males based on signs of breeding condition.

Imaging

Unlike mammals, in which the skeletal structure of the respiratory conchae are typically composed of bone (e.g., Van Valkenbergh et al. 2011), the skeleton of avian conchae is primarily cartilaginous with some regions ossified to varying degrees in different species (Bang 1971). Because cartilage is damaged or destroyed during the preparation of dry museum study skins and skeletons, the gross anatomy of avian conchae has traditionally been studied using histology of fluid-preserved specimens. Contrast-enhanced CT is a relatively new technique that enables simultaneous 3-dimensional (3D) visualization of hard and soft animal tissues (Metscher 2009). We employed this technique to visualize and measure avian conchae in fluid-preserved specimens in situ, without sectioning or dissection.

CT scans were performed at the Cornell Imaging Multiscale CT Facility ( http://www.ct.cornell.edu). Birds were removed from the iodine solution the day before scanning, and compressed air was blown through each naris for 10 s to expel ethanol droplets and improve contrast around the conchae. The bill of each bird was covered in tissue and the bird stored in a zip-lock bag until scanned. The birds were scanned in a VersaXRM-520 ZEISS Xradia microscope (Carl Zeiss X-ray Microscopy, Inc., Pleasanton, CA). Scan scale ranged from 16.3 to 20 μm.

We radiographed study skins at the Smithsonian Museum Support Center using a Kevex Microfocus X-ray source (PXS10-16W, Kevex X-Ray Products, Scotts Valley, CA) with a 6-micron focal spot along with a Varian System Flat Panel Amorphous Silicon Digital X-Ray Detector (Pax-Scan 4030R, Varian Medical Systems, Inc., Palo Alto, CA) with a 28.2 × 40.6 cm pixel area. The product was a 7.1 MB 8 bit TIFF file, captured by Viva k.03 software (Varian Medical Systems, Inc.).

Description of Anatomical Structures and Measurement

To characterize the structure and location of the conchae, nasal cavity, and external morphology, we examined 3D renderings and standardized cross-sections from CT scans and individual radiographs. CT scan measurements were taken using the 3D MPR rendering on OsiriX 5.8.5, 64-bit (Pixmeo SARL). All individual images used were screenshots taken along an axis through each bird's nasal cavity. The axis ran along the nasal septum beginning at the anterior edge of the craniofacial flexion zone and ending at the distal end of the bony nasal cavity (Appendix Figure 5). We took 10 equally spaced frontal (i.e. coronal) cross-sectional screenshots along this axis (Figure 2), starting at the caudal end (i.e. the first section was at the craniofacial flexion zone). The distance between sections averaged 0.570 mm (SE = 0.005 mm, range = 0.517–0.599) and did not differ by subspecies, sex, or structural sizes (tarsus length and wing length; all t < |1.5|, all p > 0.15).

FIGURE 2.

Frontal cross-sections 3–8 of the bill in 2 Song Sparrows, illustrating the differences between (A) M. m. atlantica and (B) M. m. melodia. Sections are ordered sequentially, from caudal-most (top left, section 3) to rostral-most (bottom right, section 8). Within each cross-section, the top is dorsal and the bottom is ventral. Note that the complexity has shifted rostral in panel (A). Cross-sections are derived from high-resolution CT scans of fluid-preserved specimens. The 4 upper left images show the scroll-like middle conchae, and the right and lower left images show the branched rostral conchae with their accessory ridges. Scale bar = 2.0 mm.

Using CT scans, we measured complexity differently in middle and rostral conchae. In each frontal cross-section of the middle concha, we counted the number of times the middle concha curled below and above the midpoint and reached the horizontal. For example, if the concha scrolled 1.5 revolutions, we assigned it a score of 3 (Appendix Figure 6). For rostral conchae, we counted how many ridges branched off the main concha body in each section, counting only ridges that were taller than they were wide (Appendix Figure 6).

From CT scans, we measured surface areas of the conchae and external bill as well as linear dimensions of the conchae, nasal capsule, and external bill. We calculated the surface area of the middle and rostral conchae by combining measurements from the 10 frontal cross-sections described earlier and the length of the conchae. First, on each of the frontal cross-sections, we measured the total linear distance along the edge of the concha that would contact respired air using ImageJ 1.48 (Rasband 2014) and a scale bar from the OsiriX measurement tool. We then sequentially averaged the measurements of adjacent screenshots and multiplied those averages by the distance between each pair of screenshots. Last, we summed those products separately for middle and rostral conchae to estimate surface area for each concha. We measured the length of the concha from the rostral to the caudal tips of each (Appendix Figure 7A and B). To quantify their overlap, we measured the distance between the distal end of the middle concha and the proximal end of the rostral concha parallel to the axis that ran from the ectethmoidal notch to the tip of the nasal cavity (Appendix Figure 7C). To measure nasal cavity size, we took width and depth measurements perpendicular to this axis, where the lateral nasal skeletal bars were visible on those planes and thus provided a consistent landmark (Appendix Figure 7D and E). External bill dimensions, including culmen length, bill width, and bill depth, were taken level with the anterior end of the nares. Total bill length was measured from the bill tip to the craniofacial flexion zone (i.e. frontonasal hinge).

From individual radiographs of dry study skins, we took linear measurements using ImageJ 1.48 (Rasband 2014), including nasal cavity depth from the highest point of the dorsal wall of the nasal cavity to the ventral wall of the nasal cavity (Appendix Figure 8) and width along the ventral wall of the nasal cavity, between the left and right maxillary processes of the nasal bone. We were unable to identify a clear landmark at the base of the bill that would allow measurements of nasal cavity length. From these dry specimens, we also measured external bill length (nalospi), width, and depth at the anterior edge of the nares using calipers (0.01 mm precision).

Analyses

Because we took CT scans and individual radiographs from different groups of specimens, we analyzed those datasets separately. Using measurements from CT scans, we tested for relationships between conchae size and nasal cavity size and between the nasal cavity and external bill morphology. Between subspecies, we compared the surface area and overlap of conchae. We corrected surface area of the conchae by dividing by nasal cavity length because the surface area was correlated with the cavity length (see results). We compared the complexity of the middle and rostral conchae and the position of the maximum complexity of both conchae, which were uncorrected for nasal cavity length. With data from radiographs, we compared nasal cavity size as a proxy for concha size.

We performed all statistics in R (R Core Team 2015). We used linear models (function lm) with a normal error distribution for all comparisons of continuous variables. We visually inspected plots of residuals from linear models and found they were normally distributed. For count data (i.e. complexity of conchae), we used the nonparametric Mann-Whitney U test (function wilcox.test). For comparisons of subspecies we present group means (x̄_atlantica and x̄_melodia), and for relationships between structures we present effect sizes (i.e. B). We also present the associated standard errors for differences, test statistics (t- or z-values), and p-values. We rejected null hypotheses with a p-value <0.05. Because we performed multiple tests of each hypothesis, which increases the probability of a type I error (falsely rejecting the null hypothesis of no difference between subspecies), we adjusted our p-values with the Holm-Bonferroni correction for each family of hypotheses (Holm 1979) using R function p.adjust.

Subspecies did not differ in structural size (wing length and tarsus length) in either the fluid-preserved or dry specimens (all t < |0.8|, all p > 0.16), consistent with previous studies of these populations (Greenberg et al. 2012, Danner and Greenberg 2015). Among the fluid-preserved specimens used for CT scans, no differences were found between subspecies in external bill dimensions (all t < |1.5|, all p > 0.17), suggesting that the sample sizes were too small to detect the differences recorded in other studies (Greenberg et al. 2012, Danner and Greenberg 2015), and also that allometry would not bias intergroup comparisons of other morphological features. Among the larger sample of study skin specimens that we radiographed, the dune endemic subspecies had wider, deeper, and longer bills at the nares, and as a result had greater bill surface areas (all t > |2.7|, all p < 0.009), consistent with previous studies (Greenberg et al. 2012, Danner and Greenberg 2015). Nasal cavity size was not related to the date the specimen was collected (both t < |1.2|, p > 0.25), indicating that collection date did not bias intergroup comparisons.

Description of Anatomical Structures

As shown in the CT images (Figure 1), 2 conchae dominate the interior space of the nasal cavity in the Song Sparrow and effectively divide the cavity into 2 connected chambers. Air inspired through the naris first enters the nasal vestibule and encounters the rostral concha, a complex structure consisting of a central plate suspended from the dorsolateral wall of the nasal cavity, with a number of ridges projecting from it at approximately right angles (usually 5 ridges in our sample). In cross-section, this concha appears as a branching structure (Figure 1 and 2). The ridges of the rostral concha interdigitate with smaller ridges that arise from the nasal septum and the lateral wall of the nasal vestibule (usually 3 from the nasal septum and 2 from the lateral wall; Figure 2). Air next enters the space dominated by the middle concha, a simpler scroll-shaped structure with an arched roof dorsally, suspended again from the dorsolateral wall of the nasal cavity. The middle concha is coiled to varying degrees but usually ~1.5 times at the maximum. After flowing over the middle concha, inspired air can progress through the pharynx and down the trachea.

The rostral concha and its accessory ridges occupy the space from the front of the nasal cavity to slightly posterior to the external naris. Toward its posterior terminus, it is overlain dorsally by the middle concha with its surrounding space or chamber. That chamber extends from just posterior to the external naris back beyond the bill into the antorbital space, posteriorly. The chamber defined by the middle concha extends all the way to the ectethmoid plate. We did not observe a third caudal concha in the CT scans of Song Sparrow heads.

The middle and rostral conchae, together with interdigitating ridges, fill the nasal cavity along both left–right and dorsoventral axes; their size is restricted by the surrounding dimensions of the bony maxilla and to a lesser extent the antorbital space (example in Figure 1). Based on measurements from CT scans, external bill depth was related to nasal cavity depth at the lateral nasal bars (B = 2.25 mm ± 0.60 SE, t₁₃ = 3.8, p = 0.002, r² = 0.52), and external bill width tended to be related to nasal cavity width (B = 1.07 mm ± 0.52 SE, t₁₃ = 2.1, p = 0.06, r² = 0.25). Along the anteroposterior axis, longer nasal cavities housed longer middle and rostral conchae (middle: B = 0.76 mm ± 0.17 SE, t₁₃ = 4.5, p < 0.002, r² = 0.61; rostral: B = 0.42 mm ± 0.19 SE, t₁₃ = 2.2, p < 0.05, r² = 0.27). Longer nasal cavities housed conchae with larger surface areas (middle: B = 8.22 mm² ± 3.77 SE, t₁₃ = 2.2, p = 0.048, r² = 0.27; rostral: B = 9.73 mm² ± 3.33 SE, t₁₃ = 2.9, p = 0.024, r² = 0.35). External bill surface area correlated with surface area of conchae (B = 0.36 mm² ± 0.14 SE, t₁₃ = 2.7, p = 0.021, r² = 0.39).

Subspecies Comparisons

Based on CT scans, after correcting for nasal cavity length, surface area of conchae was significantly larger in the dune endemic subspecies (x̄_atlantica = 71.6 mm² ± 2.64 SE, x̄_melodia = 64.7 mm² ± 3.25 SE, t₁₃ = −2.33, p < 0.01; Figure 3). Independent measurements of the middle and rostral conchae (corrected for nasal cavity length) tended to be larger in the dune endemic subspecies (middle: x̄_atlantica = 32.5 mm² ± 1.39 SE, x̄_melodia = 28.9 mm ± 1.98 SE, t₁₃ = −1.84, p = 0.09; rostral: x̄_atlantica = 39.0 mm² ± 1.52 SE, x̄_melodia = 35.8 mm ± 1.81 SE, t₁₃ = −1.89, p = 0.08). The middle concha of the dune endemic extended farther anterior than in the interior subspecies (Figure 4 vertical lines; x̄_atlantica = 5 sections ± 0 SE sections and x̄_melodia = 4.3 sections ± 0.2 SE), creating greater overlap with the rostral concha (x̄_atlantica = 2.5 mm ± 0.20 SE, x̄_melodia = 1.57 mm ± 0.32 SE, t₁₃ = −2.30, p < 0.04). The location of maximum complexity tended to be shifted anterior in the dune endemic subspecies in both the middle concha (x̄_atlantica = 3.5 sections ± 0.10 SE, x̄_melodia = 3.21 sections ± 0.11 SE, t₁₃ = −2.1, p = 0.06) and the rostral concha (x̄_atlantica = 7.75 sections ± 0.17 SE, x̄_melodia = 7.21 sections ± 0.11 SE, t₁₃ = −2.7, p = 0.04). The maximum complexity did not differ between subspecies; all individuals had a maximum of 3 internal surface features in each middle conchae, and rostral conchae had a range of 4–6 features (x̄_atlantica = 5.87 ± 0.13 SE, x̄_melodia = 5.57 ± 0.32 SE, W = 33, p = 0.46).

FIGURE 3.

Conchae surface area in relation to nasal cavity length in 2 subspecies of Song Sparrow. Points = raw data, solid lines = model predictions, dashed lines = predicted standard error.

FIGURE 4.

Count of concha complexity across 10 cross-sections of 2 subspecies of Song Sparrow. Solid lines = mean, dashed lines = standard error. The x-axis is the number of the frontal cross-section of the bill, from caudal to rostral, where 0 = the craniofacial flexion zone and 9 = the distal end of the nasal cavity. Vertical dotted lines indicate the mean of the rostral terminus of the middle concha. In M. m. atlantica, the rostral terminus was always at section 5, hence the lack of error bars.

Among study skins, radiographs showed that the dune endemic subspecies had nasal cavities that were both deeper (x̄_atlantica = 2.52 mm ± 0.038 SE, x̄_melodia = 2.40 mm ± 0.022 SE, t₄₀ = −3.5, p = 0.001; controlled for tarsus length by including tarsus length as an additive variable) and wider (x̄_atlantica = 4.08 mm ± 0.030 SE, x̄_melodia = 3.93 mm ± 0.017 SE, t₄₀ = −4.3, p < 0.001) than those of the inland subspecies.

Description of Anatomical Structures

In the Song Sparrow, the conchae together with accessory ridges fill the nasal cavity and provide surface area to condition inspired air before it enters the lungs and to recapture moisture from expired air. The conchae of the Song Sparrow exhibit both differences and similarities with conchae of other species. The structure of the nasal conchae has previously been documented in only 15 passerine families and 26 species, to our knowledge, none of which were found to have an elaborate rostral concha with interdigitating accessory ridges as found in M. melodia (Bang 1971, Yokosuka et al. 2009a, 2009b). In nearly all other birds studied, the nasal septum is free of even minor projections. The tube-nosed seabirds (Procellariiformes) are exceptions to the rule; they are the only birds in which a concha has been previously reported to arise from the nasal septum (Bang 1965). This concha is olfactory in function (Bang 1965), however, and is not associated with the rostral respiratory conchae, which are themselves greatly reduced or occluded, putatively to protect internal structures while plunge-diving (Geist 2000). Whether the elaborate respiratory structures densely filling the nasal vestibule we observed in the Song Sparrow are typical of new world sparrows is unknown.

The structure of the middle concha and lack of a caudal concha are similar to descriptions of other taxa (Bang 1971, Yokosuka et al. 2009a, 2009b). The middle conchae occupy the greater part of the bill posterior to the nares as well as much of the antorbital space, and they comprise the majority of the soft structure inside the nasal cavities. The middle concha has the typical scroll-like form found in most other birds (Bang 1971). The apparent absence of a distinct caudal olfactory concha in the Song Sparrow is consistent with reports of other passerines in which the caudal concha is described as either a mound or lacking entirely (Bang 1971, Yokosuka et al. 2009a, 2009b). The poorly developed or absent olfactory concha by itself tells us little about the olfactory abilities of the Song Sparrow. In the passerine birds examined by Bang (1971), olfactory epithelium was frequently present on the caudodorsal roof of the nasal cavity and sometimes on the nasal septum.

Subspecies Comparisons

The anatomical differences between the dune endemic M. m. atlantica and sister subspecies M. m. melodia suggest current or past divergent selection pressures on nasal conchae. First, the larger surface area of the conchae in the dune endemic subspecies would seem to allow greater water condensation during exhalation, thus presumably facilitating improved water efficiency. Geist (2000) found that functional conchae resulted in savings of 55–71% of respiratory water loss in 4 species of temperate-zone birds. The large potential water savings afforded by conchae suggests that increases or decreases in surface area may lead to significant changes in water conservation. Second, we found that the locations of maximum complexity within the conchae were more distal in the dune endemic subspecies. The functional significance of this shift is unknown. Tieleman et al. (1999) hypothesized that longer bills may allow greater cooling capacity and therefore greater potential for water recapture. A potential mechanism underlying this relationship is that longer bills position condensing surfaces distally, farther from the core of the bird, thus providing cooler surfaces, which may facilitate condensation of expired air.

To our knowledge, the structure of avian conchae has not previously been compared across a moisture gradient. Geist (2000) noted in a personal observation that 4 species within 1 family (Phasianidae) that inhabit different climate zones appear qualitatively similar, although measurements are not provided. Tieleman et al. (1999) and Sabat et al. (2006) hypothesized that species of larks (Ammomanes deserti and Galerida cristata) and populations of Rufous-collared Sparrow (Zonotrichia capensis), respectively, would have larger or more complex conchae in arid locations, which would result in greater water efficiency. They found that the taxa from arid habitats did not recover more moisture in the nasal passages, although they did not test whether the structure of the conchae varied along the moisture gradient. Similarly, Schmidt-Nielson et al. (1970) found no difference in exhaled air temperature (a proxy for water recapture in the nasal cavity) among 7 species of birds (many from different families) that inhabit deserts vs. more moderate environments, although the structures of conchae were not measured. To understand the functional anatomy of conchae, comparisons of evaporative water loss, breathing rates, and simulated airflow should focus on taxa with described differences in concha structure, such as our study populations.

Among mammals, interspecific divergence in conchae and the bony turbinates has been shown along the transition from land to salt water, suggesting adaptation to those environments. Van Valkenburgh et al. (2011) showed that aquatic and semiterrestrial mammals of saltwater environments tend to have relatively larger respiratory conchae than their terrestrial relatives. Van Valkenburgh et al. (2011) hypothesized that larger respiratory conchae have evolved to increase water and heat retention in response to the limited fresh water in marine environments and greater potential for heat loss underwater. The large respiratory conchae of elephant seals (Mirounga angustirostris) contribute to this species' ability to retain up to 92.5% of the water contained in each breath (Lester and Costa 2006).

Although larger conchae putatively improve water conservation, they may also recapture more body heat, potentially posing a challenge for animals that inhabit hot climates, such as the dune-endemic M. m. atlantica; however, the potential of respiratory conchae to conserve heat seems to be greatest at lower ambient temperatures. Geist (2000) estimated that at a cool temperature (15 °C), the conchae of 4 temperate zone bird species conserve 3.5–6.1% of the daily energy budget. Schmidt-Nielsen et al. (1970) and Murrish (1973) used physical and physiological principles to estimate that heat recapture declines at higher temperatures, when heat loss would be advantageous, suggesting that larger conchae would not be a liability in hot conditions.

The conchae and external bill of the Song Sparrow are tightly nested structures, positively correlated in size, and have related functions of heat and water balance (Greenberg et al. 2012, Danner and Greenberg 2015), suggesting they are phenotypically integrated (Pigliucci 2003) and evolve in tandem. We hypothesize that in our system, the larger external bill of the dune population has evolved, at least in part, to accommodate larger conchae. Because the dune birds experience hot, dry summer conditions but relatively mild winters (Danner et al. 2016), we expect the season of critical thermal stress to be summer (Danner and Greenberg 2015). As a result, we expect selection to favor both larger conchae and external bill surface area in the dune-endemic subspecies. Regarding the interior population, Danner and Greenberg (2015) hypothesized that winter is the season of critical thermal stress because winters are cold, and, although summers are hot, moisture is plentiful. In interior populations, we therefore expect selection for smaller bill surface area in the winter to minimize heat loss. Larger conchae would theoretically improve heat recapture in interior winters, however, thus raising the possibility that the smaller bills constrain the size of conchae in that population and that the season of critical thermal stress may differ from organ to organ (i.e. external bill vs. conchae).

Although we generally hypothesize that the hot/dry and cool/moist climates in our system select for positively correlated sizes between concha and external bill size, other climates may select different relative sizes of these structures. For example, in cold and dry systems we might expect birds to minimize bill surface area to reduce heat loss yet maximize concha size and complexity to recapture water and heat. In hot and moist environments, however, we expect small conchae because there is a lower premium on water, yet larger bill surface area for heat dissipation because heat loss through evaporative water loss is less effective in humid environments. In climates with ambient temperatures that frequently rise above the body temperature of the organism, we expect smaller bills to reduce heat gain from the environment (Greenberg and Danner 2012). Further, because the respiratory conchae are expected to heat up to body temperature or above, they would not function as countercurrent heat exchangers at temperatures above body temperature and would therefore not recapture moisture; the type of selection on conchae, or lack thereof, is unclear in this situation. Last, adaptation of conchae to specific climates may include the evolution of complexity independent of size. Note that many variables likely influence the evolution of concha size and external bill morphology, and thus these patterns may not be found in all taxa.

Our study reveals that respiratory conchae of the Song Sparrow are more complex in structure than expected and provides the first evidence that conchae can vary in size and structure within a species, between habitats. Contrast-enhanced CT scanning enables the morphology of these cryptic structures to be conveniently visualized and compared, which should facilitate further study of avian conchae in ecological and evolutionary contexts. To better understand function of concha structure and size, researchers could pair anatomical information with measured or expected abilities of air conditioning under various climatic conditions. In addition, our results with the Song Sparrow illustrate that much work remains to be done to simply characterize nasal conchae among avian taxa, particularly among passerines.