The middle ear of the pink fairy armadillo Chlamyphorus truncatus (Xenarthra, Cingulata, Chlamyphoridae): comparison with armadillo relatives using computed tomography

Abstract

The pink fairy armadillo Chlamyphorus truncatus is the smallest extant armadillo and one of the least‐known fossorial mammals. The aim of this study was to establish if its middle ear is specially adapted to the subterranean environment, through comparison with more epigeic relatives of the groups Euphractinae (Chaetophractus villosus, Chaetophractus vellerosus, Zaedyus pichiy) and Dasypodinae (Dasypus hybridus). We examined the middle ears using micro‐computed tomography and subsequent three‐dimensional reconstructions. D. hybridus has a relatively small middle ear cavity, an incomplete bulla and ‘ancestral’ ossicular morphology. The other species, including Chlamyphorus, have fully ossified bullae and middle ear ossicles, with a morphology between ‘transitional’ and ‘freely mobile’, but in all armadillos the malleus retains a long anterior process. Unusual features of armadillo ears include the lack of a pedicellate lenticular apophysis and the presence, in some species, of an element of Paaw within the stapedius muscle. In common with many subterranean mammals, Chlamyphorus has a relatively flattened malleo‐incudal articulation and appears to lack a functional tensor tympani muscle. Its middle ear cavity is not unusually enlarged, and its middle ear ossicles seem less robust than those of the other armadillos studied. In comparison with the euphractines, there is no reason to believe that the middle ear of this species is specially adapted to the subterranean environment; some aspects may even be indicative of degeneration. The screaming hairy armadillo, Chaetophractus vellerosus, has the most voluminous middle ear in both relative and absolute terms. Its hypertrophied middle ear cavity likely represents an adaptation to low‐frequency hearing in arid rather than subterranean conditions.

Three‐dimensional reconstructions of right middle ear structures of Chlamyphorus truncatus (A), Chaetophractus villosus (B); Chaetophractus vellerosus (C, D), Zaedyus pichiy (E) and Dasypus hybridus (F). AP, anterior process; ECTB, ectotympanic bone; EP, element of Paaw; ER, epitympanic recess; I, incus; M, malleus; MC, mastoid cavity; S, stapes; SFP, stapes footplate; SM, stapedius muscle; TC, tympanic cavity; TT, tensor tympani muscle.

Introduction

The Xenarthra comprise an ancient group of South American mammals, the most successful of the placental assemblages that dispersed to the north during the Great American Biotic Interchange of the late Tertiary (McDonald, 2005). Despite their great diversity in the past, there are relatively few extant armadillos, sloths and anteaters. These animals are widely distributed in South and Central America with a single species, the nine‐banded armadillo (Dasypus novemcinctus), reaching southern North America. The position and phylogenetic relationships of Xenarthra are controversial and under constant revision: at present they are considered as one of the four major clades of placental mammals (Madsen et al. 2001; Murphy et al. 2001) and the sister group of Afrotheria within Atlantogenata (Tarver et al. 2016).

Armadillos (Cingulata) are the most speciose xenarthrans today. Based on molecular phylogenetic analyses, Gibb et al. (2016) proposed assigning all the extant species to two distinct families, Dasypodidae and Chlamyphoridae. The former contains only one subfamily (Dasypodinae), represented by the single genus Dasypus (long‐nosed armadillos). The latter includes eight genera grouped into three subfamilies: Chlamyphorinae (fairy armadillos), Euphractinae (hairy armadillos) and Tolypeutinae (giant, three‐banded and naked‐tailed armadillos). Delsuc et al. (2012) proposed that the fossorial lifestyle of fairy armadillos probably evolved as a response to the Oligocene aridification that occurred in South America, after their divergence from the Tolypeutinae around 32 million years ago.

The possession of a flexible carapace formed by dermal bones is not the only bizarre characteristic of armadillos. Other peculiarities are homodonty, absence of enamel in adults, dental aberrancies (Sidorkewicj & Casanave, 2013), a striking plasticity of some of their organ systems (Casanave & Galíndez, 2008), an obligate monozygotic polyembryony in Dasypus (Loughry et al. 1998) and relatively low metabolic rates (McNab, 1980). Almost all members of the group will burrow to some extent, although they differ in their digging habits (Milne et al. 2009; Galliari, 2014). Vizcaíno et al. (1999) classified armadillos into three categories, later modified by Milne et al. (2009): (i) non‐diggers (mainly cursorial species: Tolypeutes); (ii) generalised diggers (Dasypodinae and Euphractinae); and (iii) specialised diggers (giant armadillo Priodontes maximus, naked‐tailed armadillos of the genus Cabassous, and the chlamyphorines Chlamyphorus truncatus and Calyptophractus retusus).

The pink fairy armadillo C. truncatus, known also as pichiciego menor or pichiciego pampeano, is the smallest extant armadillo. It is usually considered a strictly subterranean species (Borghi et al. 2011; Delsuc et al. 2012; Torres et al. 2015), but it lacks the extreme anatomical adaptations to a subterranean environment found in, for example, some talpid moles, mole‐rats and golden moles. Although it feeds mainly underground, it also consumes above‐ground items (Meritt, 1985) and is said to leave its burrows occasionally (Minoprio, 1945; Rood, 1970). A more appropriate term for C. truncatus might be ‘fossorial’ as defined by Lange et al. (2004), i.e. having habits intermediate between generalised epigeic and strictly subterranean forms. Chlamyphorus is endemic to central Argentina (Wetzel et al. 2007), where it inhabits sandy plains, dunes and scrubland. Categorised as a sand‐swimmer by Borghi et al. (2002), it is rarely seen because of its nocturnal and subterranean habits, and this makes its study particularly difficult. Although it has been generally accepted that this species has declined in distribution and abundance in recent years (Superina, 2006; Aguiar & Fonseca, 2008; Ojeda et al. 2012), Borghi et al. (2011) found that populations persist along most of their original range and added new records outside the historical distribution map. However, there is an almost total lack of biological data on this species, one reason why it is currently listed as Data Deficient in the IUCN Red List of Threatened Species (Superina et al. 2014). There is an urgent need to fill gaps in our knowledge of these enigmatic creatures.

The hearing of subterranean and fossorial mammals has attracted considerable interest owing to the unusual acoustic environment underground. Several studies have identified common middle ear features in species inhabiting underground ecotopes (e.g. Burda et al. 1989, 1992; Mason, 2001, 2013; Begall & Burda, 2006). The malleo‐incudal complex typically has a ‘freely mobile’ morphology, to use Fleischer’s (1978) terminology, characterised by ossicles with relatively large heads, a manubrium of the malleus roughly perpendicular to the ‘anatomical axis’ extending between anterior process of the malleus and short process of the incus, and relatively loose attachments of these two ossicular processes with the skull. Other characteristics commonly found among subterranean mammals are reduced or absent middle ear muscles, stapedial arteries (where present) contained within bony tubes, tympanic membranes without a pars flaccida and relatively large stapes footplates (Burda et al. 1992; Mason, 2001, 2003, 2004, 2013, 2015; Begall & Burda, 2006; Begall et al. 2007; Mason et al. 2010, 2016). Low‐frequency sounds have been found to propagate better than higher frequencies in subterranean tunnels (Heth et al. 1986), and it has duly been suggested that at least some of these ‘subterranean’ ear characteristics improve the transmission of low‐frequency sound to the cochlea (Burda et al. 1989, 1992). However, the most obvious anatomical feature that one would expect in a ‘low‐frequency’ middle ear would be a capacious cavity, serving to increase acoustic compliance (Mason, 2016a). Those subterranean mammals which have been investigated so far do not have significantly larger middle ear cavities than non‐fossorial taxa (Mason, 2001). Accordingly, behavioural audiograms from subterranean rodents show that, while they tend to have hearing restricted to low frequencies, it is not unusually acute at those frequencies (Heffner & Heffner, 1990, 1992, 1993; Brückmann & Burda, 1997; Gerhardt et al. 2017).

In order to assess whether ear structures in subterranean species represent true adaptations towards hearing underground, they must be compared with those of close, terrestrial relatives. Among talpid moles (Eulipotyphla; Talpidae), ‘subterranean’ middle ear characteristics are found in the more exclusively subterranean species, and are clearly derived in comparison with, for example, shrews and shrew‐moles (Mason, 2006). The middle ears of spalacid mole‐rats (Rodentia; Spalacidae) also seem to be derived in comparison with those of terrestrial muroid rodents (Mason et al. 2010). In other subterranean rodents such as bathyergid mole‐rats, Ctenomys and Spalacopus, however, similar anatomical features of the middle ear appear to be retained, primitive characteristics of the Ctenohystrica group to which these animals belong (Mason, 2004, 2016a; Begall & Burda, 2006; Argyle & Mason, 2008).

There is only limited information available regarding the middle ear anatomy of the pink fairy armadillo (Hyrtl, 1845; Roig, 1972; Fleischer, 1973; Segall, 1976; Patterson et al. 1989, 1992). Segall (1976) concluded that the ear of Chlamyphorus is ‘influenced by its fossorial life’, but a comprehensive, comparative account is lacking. We present the first micro‐computed tomography (CT) reconstructions of the middle ear of C. truncatus, and compare these with those of more epigeic armadillos of the groups Euphractinae and Dasypodinae, coming from the same geographical area. The purpose was to establish if the middle ear of C. truncatus does indeed show signs of being specially adapted to the subterranean environment.

Materials and methods

Preparation of samples

Thirteen skulls of adult armadillos belonging to the collection of the Cátedra de Anatomía Comparada and INBIOSUR‐CONICET (UNS) were used. Species were C. truncatus (pink fairy armadillo; n = 3), Chaetophractus villosus (large hairy armadillo; n = 4), Chaetophractus vellerosus (screaming hairy armadillo; n = 1), Zaedyus pichiy (pichi; n = 3) and Dasypus hybridus (southern long‐nosed armadillo; n = 2). All animals came from the province of Buenos Aires, Argentina. The specimens of C. truncatus had been deposited in the collection prior to the study, and originated from private donations. The other material came from animals found dead but with the skulls in good condition; they were collected by permission of the Ministry of Agroindustria of the Province of Buenos Aires, Flora and Fauna Division (Regulation No 77, 28/09/2017; Exp. 22500‐41961/17). Specimen data are presented in Table 1.

Table 1.

Details of specimens used in this study with indication of sex, total skull length and what was scanned.

Family (subfamily)	Species	Specimen code	Body mass, g	Sex	TSL, mm	Scanned
Chlamyphoridae (Chlamyphorinae)	Chlamyphorus truncatus	UNSCTMA1	99 ± 5 (n = 4)	Male	39.34	P
		UNSCTSI1		Unknown	36.12	P
		UNSCTSI2		Unknown	36.61	S
Chlamyphoridae (Euphractinae)	Chaetophractus villosus	UNSCVIMA87	3285 ± 726 (n = 21)	Male	94.49	E
		UNSCVIMA91		Male	92.25	E
		UNSCVIHA83		Female	94.57	P, S
		UNSCVIHA89		Female	93.70	E
	Chaetophractus vellerosus	UNSCVEHA80	1040 ± 135 (n = 11)	Female	69.02	E, P, S
	Zaedyus pichiy	UNSZPMA52	986 ± 186 (n = 13)	Male	64.42	S
		UNSZPHA55		Female	65.48	E, S
		UNSZPSI4		Unknown	66.03	P
Dasypodidae (Dasypodinae)	Dasypus hybridus	UNSDHHA2	1775 ± 456 (n = 8)	Female	72.77	S
		UNSDHHA4		Female	76.13	P

E, isolated temporal bone/ear region; P, posterior skull; S, whole skull; TSL, total skull length. Body masses of adults (mixed sexes) of comparable skull sizes and coming from the same geographical area are given (mean ± standard deviation), with an indication of the sample size (n) with which that information was obtained (data from Sidorkewicj, Basso & Casanave, unpublished).

After removal, the heads were preserved frozen until use. They were then defrosted and subjected to repeated boiling periods (15–20 min each) within a mixture of water and biological laundry detergent, until the muscles detached from the bone. Between boiling periods, the bone was allowed to dry completely and the cleaning was continued with the help of dissection implements. Although the external soft tissues of the skull were largely removed by this cleaning procedure, the middle ear muscles were allowed to dry in situ together with, in Dasypus, some of the material holding the ectotympanic in place. Total skull length (TSL; anterior edge of the premaxilla to the most posterior point of the nuchal crest) was measured in the cleaned skulls by means of digital callipers (0.01 mm); measurements are presented in Table 1.

Micro‐computed tomography and reconstruction

Micro‐CT scans were made of intact skulls of two specimens of Z. pichiy, and one specimen of each of the other species considered in this study. To give enhanced detail of the middle ear structures, scans were also made of the posterior skull only, and in some cases dissected‐out temporal regions, each of which included one auditory bulla (Table 1). Samples were placed on radiotranslucent material to prevent movements during scanning. Scans were made using a Nikon XT H 225 micro‐CT scanner. The settings used were 125–130 kV and 120–130 μA. The images were reconstructed from 1080 projections, each with an exposure time of 1000 ms and two frames averaged per projection. CT AGENT XT 3.1.9 and CT PRO 3D XT 3.1.9 (Nikon Metrology, 2004–2013) were used to process the scan data. Cubic voxel side‐lengths were 13.8–50.9 μm (see Appendix S1 for scan details).

To facilitate image processing, exported 16‐bit tiff files were converted to 8‐bit jpg files using Adobe Photoshop CS 8.0 (Adobe Systems Inc. 2003). Most of the three‐dimensional reconstructions were made using Stradwin 5.4 (Graham Treece, Andrew Gee & Richard Prager, 2018). Within this program, the boundaries of structures of interest were identified and outlined in order to create the reconstructions. The outlining of ossicles involved automated thresholding followed by manual correction, but where boundaries were less distinct, all outlining had to be performed manually. For larger structures such as cavities, structures were outlined in a subset of the available tomograms, the software interpolating in‐between. For smaller structures, such as ossicles, a higher proportion of the available tomograms was used. Reconstructions of the whole skulls and tympanic rings were made using MicroView 2.5.0 (Parallax Innovations Inc., 2019). Images of right or left ears were laterally inverted where necessary, to facilitate comparison in the composite figures.

Morphofunctional parameters

The parameters measured from the three‐dimensional reconstructions were volumes of the middle ear cavities and ossicles, lever arms of the malleus and incus, and areas of the tympanic membrane and of the stapedial footplate. Although all CT scans were examined, measurements of middle ear structures were made only from the higher‐resolution scans of the temporal bones or posterior skulls. If both ears had been scanned, right‐side measurements only were recorded unless that ear was damaged.

The volumes of the tympanic cavity and associated subcavities were calculated from reconstructions of their boundaries. The position of the tympanic membrane, which forms part of the boundary of the tympanic cavity, was in each case estimated from the positions of the bony tympanic annulus and the manubrium mallei. In order to calculate air‐space volumes, middle ear ossicle and muscle volumes were subtracted from the overall cavity volumes. Because there is no very clear demarcation between the confluent tympanic cavity and epitympanic recess, the division that was made between them was somewhat arbitrary. As such, their relative volumes must be considered as approximations only.

The line between anterior process of the malleus and short process of the incus has been referred to as the ‘anatomical axis’ (Lavender et al. 2011), used as an estimate of where the ossicular rotatory axis might be, at low frequencies. Because the anterior process is very long in armadillos and in articulation with the skull along most of this length, the thin, proximal part of the process, where it meets the transversal lamina, was considered to be the most likely point of flexibility about which the ossicles would rotate. Stradwin was therefore used to obtain the spatial coordinates (x, y, z) of a point in this position on the anterior process (A), and another at the short process of the incus. These were used to calculate the equation of the line joining both points. The coordinates of a point at the tip of the manubrium (B) were also obtained. To calculate the malleus lever arm (ML) two vectors were considered, w with origin in A and end in B, and the vector v extending along the anatomical axis with the same origin as w. The orthogonal projection of w on v was

$$|w|\cos \alpha =\frac{(w,v)}{|v|},$$

where $〈w,v〉$ is the scalar product between vectors, |w| and |v| are the moduli of vectors w and v respectively, and α is the angle between them. Then the distance from the rotatory axis to the point B, which represents ML, was calculated as:

$$|\frac{(w,v)}{|v|}\frac{v}{|v|}-w|$$

The same procedure was used to obtain the incus lever arm (IL), as the perpendicular distance from the anatomical axis to the centre of the incudal articulation facet for the stapes. The anatomical lever ratio (LR) was then calculated as ML/IL. All calculations were programmed using Excel.

The tympanic membrane area (TA) was obtained from the MicroView reconstructions of the bony tympanic ring. Each reconstruction was oriented in MicroView such that the perimeter of the tympanic ring was in the plane of the computer screen, and then the two‐dimensional image (screenshot) was exported to ImageJ 1.52e (National Institutes of Health, USA). Using the automated measure function of the software and an appropriate scaling factor, the TA was estimated as a flat surface. The same procedure was used to estimate the stapedial footplate area (FA), from the medial view of the reconstructed footplate. The anatomical area ratio AR was then calculated as TA/FA.

To enable fairer comparisons among species, the parameters were considered relative to the TSL, used as an index of cranial size. Body mass was not used because, besides being subject to high intra‐specific variation depending on sex, age, nutritional and health status, etc., individual values were not available for our specimens. To take into account the allometric scaling of middle ear structures expected among mammals (Nummela, 1995; Mason, 2001), least‐squares linear regressions were performed for the relationship of each parameter with TSL on log‐transformed data. Deviations of the slope coefficient b from the theoretical value of isometric growth (b = 3, 2 or 1 for volumes, areas and linear measurements against TSL, respectively) were examined by t‐test using Excel. Since no relationship is expected between a ratio and skull length, the area and LRs of all the species were directly compared. This simple approach has obvious shortcomings: the sample size was small, the numbers of specimens of each species unequal and the relationships among species were not taken into account. For these reasons, the regression relationships were used only to identify species which clearly had relatively large or small ear structures, in comparison with the general trend.

Results

A detailed description of the external bullar anatomy of the Cingulata is found in Patterson et al. (1989); therefore, we will only briefly mention the morphological characteristics of the bulla, concentrating instead on the ossicles and other internal features that are significant from a functional point of view.

The middle ear of Chlamyphorus truncatus

The auditory bulla is completely ossified, well developed, markedly swollen and clearly demarcated from the basicranium (Fig. 1). It is roughly ovoid in ventral view and its long axis is oriented approximately 45° relative to the sagittal plane of the skull. It was not possible to distinguish the boundaries between its component bones in our adult specimens. The tympanic membrane is supported by a complete, bony tympanic ring (Fig. 2D). The bony rim is almost circular except for a rounded embayment in the epitympanic region, just dorsal to the lateral process of the malleus. This notch, which may hold a pars flaccida, represented on average 8% of the total area of the membrane contained within the rim (Table 2). The lateral part of the bulla contains the recessus meatus, which abruptly narrows to form the external opening without the interposition of a bony tube. The bullar walls are not pneumatised (Fig. 3A).

Figure 1.

3D reconstructions of the skulls of armadillos in ventral view (Chlamyphorus truncatus: UNSCTSI2; Chaetophractus villosus: UNSCVIHA83; Chaetophractus vellerosus: UNSCVEHA80; Zaedyus pichiy: UNSZPHA55; Dasypus hybridus: UNSDHHA2). The right bulla, including its lateral extension as the bony covering of the recessus meatus, is shaded in red in each case. In Dasypus, which lacks a complete bulla, the ectotympanic and associated bony elements are shaded. Enlargements of the right ear regions are shown for C. truncatus and D. hybridus. The scanned specimen of Chaetophractus villosus had a damaged rostrum, while that of C. truncatus retained mandibles. [Color figure can be viewed at https://www.wileyonlinelibrary.com]

Figure 2.

Three‐dimensional reconstructions of the right tympanic ring, malleus and incus in five species of armadillos, from approximately medial views. Positions of the tympanic membranes are indicated by light brown shading. (A) Chaetophractus vellerosus (UNSCVEHA80); (B) Chaetophractus villosus (UNSCVIMA87); (C) Zaedyus pichiy (UNSZPHA55); (D) Chlamyphorus truncatus (UNSCTSI1); E Dasypus hybridus (UNSDHHA4). The scale bar represents 5 mm. [Color figure can be viewed at https://www.wileyonlinelibrary.com]

Table 2.

Measurements made from the middle ear reconstructions in armadillos.

Species	Malleus volume, mm3	Incus volume, mm3	Stapes volume, mm3	Tympanic membrane pars tensa area, mm2	Stapes footplate area, mm2	Area ratio	Malleus lever arm, mm	Incus lever arm, mm	Lever ratio
Chlamyphorus truncatus (n = 2)	0.42 ± 0.03	0.29 ± 0.01	0.05 ± 0.01	11.32 ± 0.40	0.52 ± 0.01	21.79 ± 1.35	2.62 ± 0.07	1.20 ± 0.01	2.19 ± 0.05
Chaetophractus villosus (n = 4, except where specified)	5.78 ± 0.30	4.45 ± 0.37	0.18 ± 0.02 (n = 2)	56.39 ± 4.22	1.15 ± 0.05	49.29 ± 3.68	5.37 ± 0.48 (n = 2)	2.33 ± 0.24 (n = 2)	2.31 ± 0.03 (n = 2)
Chaetophractus vellerosus (n = 1)	4.43	3.47	0.20	44.33	1.37	32.36	5.19	1.84	2.82
Zaedyus pichiy (n = 3)	2.65 ± 0.23	1.80 ± 0.18	0.18 ± 0.07	29.92 ± 0.82	0.85 ± 0.13	35.78 ± 4.44	3.77 ± 0.29	1.80 ± 0.04	2.09 ± 0.19
Dasypus hybridus (n = 1)	1.12	0.41	0.05	10.17	0.34	29.91	2.25	1.12	2.01

Figure 3.

Transverse micro‐CT scan slices of the right side of the skull of (a) Chlamyphorus truncatus (specimen UNSCTMA1); (b) Chaetophractus villosus (UNSCVIHA83); (c and d) Chaetophractus vellerosus (UNSCVEHA80, d being a slice caudal to the external acoustic meatus); (e) Zaedyus pichiy (UNSZPMA52); (f) Dasypus hybridus (UNSDHHA4). C, cochlea; EAM, external acoustic meatus; ECTB, ectotympanic bone; ENTB, entotympanic bone; ER, epitympanic recess; IB, incus body; MH, malleus head; MM, manubrium of the malleus; P, cochlear promontory; RM, recessus meatus; RW, round window; S, stapes; TC, tympanic cavity; TM, tympanic membrane. The scale bar represents 5 mm.

Medial to the tympanic ring, the middle ear cavity is divided into two widely communicating subcavities: a large tympanic cavity and a smaller epitympanic recess (Table 3; Figs 3A and 4A). The latter, which represents 13% of the total middle ear cavity volume, houses the articulated malleus head and incudal body. It is closed laterally by a bony wall. The epitympanic recess is irregularly shaped and obliquely directed from dorsolateral to ventromedial. Below this, the cochlear promontory protrudes into the tympanic cavity, elongated in the rostrocaudal direction. The oval window is caudodorsal to the promontory and the round window caudal, the two separated by a narrow bridge of bone.

Table 3.

Total volumes of the middle ear cavity and of its subcavities, obtained from computed tomography reconstructions.

Species	TSL, mm	MEC volume, mm3	TC volume, mm3	ER volume, mm3	MC volume, mm3
Chlamyphorus truncatus (n = 2)	37.73 ± 2.28	38.64 ± 3.46	33.44 ± 3.27	5.20 ± 0.20	–
Chaetophractus villosus (n = 4)	93.75 ± 1.08	497.87 ± 71.61	407.22 ± 69.60	90.65 ± 15.58	–
Chaetophractus vellerosus (n = 1)	69.02	821.02	409.56	337.22	74.24
Zaedyus pichiy (n = 3)	65.31 ± 0.82	193.78 ± 6.97	155.11 ± 9.67	38.67 ± 11.27	–
Dasypus hybridus (n = 1)	76.13	35.07	28.68	6.39	–

ER: epitympanic recess; MC: mastoid cavity; MEC, middle ear cavity; TC, tympanic cavity; TSL, total skull length (mm). For Chlamyphorus truncatus, Chaetophractus villosus and Zaedyus pichiy, mean values (± SD) are presented.

Figure 4.

Stradwin reconstructions of right middle ear structures seen from roughly lateral views. (A) Chlamyphorus truncatus (UNSCTMA1); (B) Chaetophractus villosus (UNSCVIMA87); (C) Chaetophractus vellerosus (UNSCVEHA80); (D) detail of the middle ear ossicles and associated tissues in Chaetophractus vellerosus, from a medial view; (E) Zaedyus pichiy (UNSZPHA55); (F) Dasypus hybridus (UNSDHHA4). The walls of the cavities are shown semitranslucent to reveal the internal structures. The ectotympanic bone is shown only in D. hybridus. Not to scale. AP, anterior process; ECTB, ectotympanic bone; EP, element of Paaw; ER, epitympanic recess; I, incus; M, malleus; MC, mastoid cavity; S, stapes; SFP, stapes footplate; SM, stapedius muscle; TC, tympanic cavity; TT, tensor tympani muscle.

The middle ear ossicles are formed entirely from compact bone. The malleus and incus were not fused together, the articulation being easily distinguishable from a medial view of the ossicular reconstructions as a slightly sinuous line (Fig. 5A). The malleus head is small and essentially hemispherical, with a slight mediolateral compression. Its articular surface is oval in shape from a posterior view, and relatively shallow, the dorsomedial and ventrolateral facets not being clearly divided from each other. The anterior process is long and narrow. It curves ventrally, expanding slightly as it does so, and is fused to the ectotympanic only at the tip, with most of the process being attached only by soft tissue to the surrounding skull. There is a conspicuous perforation between the malleus head, which is excavated rostrally, and the base of the anterior process, perhaps for the passage of the chorda tympani nerve. The head continues in a short neck that retains the mediolateral compression in its first half, but then twists so that its terminal half is flattened rostrocaudally. In one of the two specimens (UNSCTMA1), we observed what could represent a markedly reduced transversal lamina between the neck and the anterior process. The manubrium is moderately long and forms an approximately 60° angle with the anatomical axis. The lateral process is rounded and not prominent, developing into a narrow margin inserting into the tympanic membrane. This inserting margin is somewhat irregularly shaped. The proximal part of the manubrium has a marked rostrocaudal compression where it becomes very thin. Although a strand of soft tissue was found to cross the middle ear cavity to insert on the manubrium of the malleus in the appropriate position, no trace of a tensor tympani muscle belly could be identified, and there is no muscular process on the malleus.

Figure 5.

Stradwin reconstructions of left middle ear ossicles of armadillos (articulated malleus and incus in approximately medial view; stapes in dorsal view). (A) Chlamyphorus truncatus (UNSCTSI1); (B) Chaetophractus villosus (UNSCVIMA87); (C) Chaetophractus vellerosus (UNSCVEHA80); (D) Zaedyus pichiy (UNSZPHA55); (E) Dasypus hybridus (UNSDHHA4). The scale bars represent 2 mm. AC, anterior crus of stapes; AP, anterior process of malleus; BI, body of incus; HM, head of malleus; LP, long process of incus; MM, manubrium of malleus; MPS, muscular process of stapes; NM, neck of malleus; NS, neck of stapes; PC, posterior crus of stapes; SFP, stapes footplate; SP, short process of incus; TL, transversal lamina of malleus.

The incus body is wide in dorsoventral direction, defining a triangular shape that has a mediolateral compression similar to that of the malleus head (Fig. 5A). The short process is sturdy and conical. The long process is wide and turns slightly medially, ending at the level of the proximal third of the manubrium. It becomes thin and spatulate distally, directly forming the articulation surface with the stapes head. The incudo‐stapedial joint is therefore based on a simple contact between the flattened terminal end of the long process and the head of the stapes, without an intervening lenticular apophysis (Fig. 6A).

Figure 6.

Stradwin reconstructions of the left incudo‐stapedial joints of armadillos, seen from rostromedially. (A) Chlamyphorus truncatus (specimen UNSCTSI1); (B) Chaetophractus villosus (UNSCVIMA87); (C) Chaetophractus vellerosus (UNSCVEHA80); (D) Zaedyus pichiy (UNSZPMA55: the stapes has become disarticulated from the incus in this specimen); (E) Dasypus hybridus (UNSDHHA4). Only the distal end of the long process of the incus and the stapes are shown. Not to scale.

The stapes is typically bicrurate (Fig. 5A). The oval head narrows to form a long and flattened neck region. This neck appears perforated in one ear of UNSCTSI1 (Figs 5A and 6A) and is broken on one side of UNSCTMA1, attesting to how thin the bone is here. There is a well‐developed muscular process; a stapedius muscle was identifiable from the CT scans but no element of Paaw was visible within it. Between neck and footplate, the stapedial crura diverge to surround a sizeable intercrural foramen, but no artery passes through this. Both stapedial crura are made of flat, thin bone, lacking internal sulci. The anterior crus is wider than the posterior one. The insertions of the crura leave around a quarter of the footplate free on each side. The flattened vestibular surface of the footplate resembles the sole of a shoe in general shape, wider caudally (Fig. 6A). It is made of thin bone, with only very slight thickening around its edge.

The middle ears in euphractine armadillos

The limits of the auditory bullae of the euphractine species (Chaetophractus villosus, Chaetophractus vellerosus and Z. pichiy) with respect to the basicranium are more irregular and less clear than in Chlamyphorus. These three taxa are characterised by a lateral extension of the bulla into a long, bony external auditory meatus, extending dorsolaterally from the recessus meatus (Figs 1 and 3B,C,E). The tympanic rings are approximately circular in these species (Fig. 2A–C). The single adult specimen of Chaetophractus vellerosus had no epitympanic notch at all. The specimens of Z. pichiy had visible sutures where the two limbs of the ectotympanic bone converged dorsally, but there was little if any notch in this position. Chaetophractus villosus had a larger epitympanic notch, but not as prominent as in Chlamyphorus. The entotympanic bones of Chaetophractus villosus and Z. pichiy, which contribute to the medial walls of the bullae, have internal cavities that are somewhat larger in the latter, but these do not represent pneumatisations from the middle ear cavity and are presumably marrow spaces (Fig. 3B,E).

In Chaetophractus villosus and Z. pichiy, the structure of the middle ear cavity is simple and, as in the pink fairy armadillo, it can be divided into tympanic cavity and epitympanic recess (Figs 3 and 4B,E). Relative to the total middle ear volume, the epitympanic recesses in these animals are similar in size to what was found in C. truncatus, representing on average 18% (Chaetophractus villosus) and 20% (Z. pichiy) of the total cavity volume. In Chaetophractus vellerosus, there are marked differences in the structure and size of the middle ear cavity (Figs 3C,D and 4C). Although it is a species of comparable body mass and skull size to Z. pichiy (Table 1), its middle ear cavity volume is on average more than four times larger. The epitympanic recess is markedly expanded dorsally, representing 41% of the total middle ear cavity volume (Table 3). Although there are low ridges projecting from its walls, it essentially represents one large cavity. Between the epitympanic recess and the tympanic cavity, just caudal to the incus, there is a largely separate compartment interpreted as a mastoid cavity, which represents 9% of the total middle ear cavity volume (Table 3; Fig. 4C). Despite the great development of the middle ear cavity in this species, there is no contact between left and right cavities in the sagittal plane of the skull.

The middle ear ossicles of these euphractine armadillos have the same general characteristics as in the pink fairy armadillo, although some differences were observed (Fig. 5B–D). The malleo‐incudal articulations are less flattened, the articulation facets on malleus and incus being more sharply demarcated from each other. The mallei differ in head shape: more robust and prominent rostrally in Chaetophractus species than in Chlamyphorus, intermediate in Z. pichiy. Small transversal laminae are present in the three taxa, this structure being least developed in Chaetophractus vellerosus (Fig. 5C). The lateral processes were more prominent in these species than in Chlamyphorus. Even though there were no distinct muscular processes, only very slight thickenings at the bases of the manubria, the scans revealed the presence of a tensor tympani muscle inserting on the malleus in at least one specimen of every species. In Chaetophractus vellerosus and Z. pichiy there is only a limited area of synostosis between the distal part of the anterior process and the ectotympanic, as in Chlamyphorus. In Chaetophractus villosus there is no fusion at all and the connection seems to occur entirely through fibrous tissue; this may account for the ease with which the malleus is dislodged and found loose inside the bulla in clean skulls of this species (APB: pers. obs.). The incus is similar in the three euphractines, although in both Chaetophractus species the long processes bend a little more medially. There was no lenticular apophysis in any species (Fig. 6B‐‐D). The stapedes, although having the same basic bicrurate morphology, differ in the robustness of their crura, resulting in intercrural foramina of different relative sizes (biggest in Chaetophractus villosus, smallest in Z. pichiy: Fig. 4). These differences account, at least partly, for the interspecific variation in the relative volumes of the stapedes (see below).

A stapedius muscle was identified in at least one specimen of each species. A small, bony element of Paaw was present within the stapedius muscle of Chaetophractus vellerosus (Fig. 4D) and some specimens of Chaetophractus villosus (UNSCVIHA83, one side only, and UNSCVIMA91), but it was never in direct contact with the stapes. No trace of an element of Paaw was detected in Z. pichiy.

The middle ear of Dasypus

The dasypodine armadillo D. hybridus has a middle ear conformation that is quite different from those of the other species studied (Fig. 1). The ectotympanic bone forms an incomplete ring, attached to the rest of the skull only by soft tissues (Figs 2E and 3F), which means that there is no complete, bony auditory bulla and no development of a bony external auditory meatus. We did not find cartilage in the ventral wall of the tympanic cavity, but several isolated bony fragments were found between ectotympanic and cochlear promontory in both specimens examined, bilaterally, which could correspond to entotympanic elements (Figs 1 and 3F). Because the ectotympanic crura do not closely converge and there was no bony lateral wall to the epitympanic recess, it was impossible to ascertain the boundaries of any pars flaccida. The TA enclosed within the ectotympanic, which is expected to comprise the pars tensa alone, is similar to that of the much smaller C. truncatus (Table 2).

Although the middle ear cavity walls in Dasypus are partially composed of soft tissue, this had dried in place and could be discerned in the CT scans, allowing us to measure the cavity volume. The components of the middle ear cavity are comparable in absolute volume with those of the pink fairy armadillo (Table 3; Figs 3F and 4F). The contributions of the tympanic cavity and epitympanic recess to the total middle ear cavity volume were 82% and 18%, respectively, markedly different from what was observed in Chaetophractus vellerosus, but similar to the other species (Table 3).

The malleus and incus are both substantially different from those of the other taxa (Fig. 5E). The malleus head is small, and its articular surface takes the form of a deep notch. The two facets meet at an acute angle. A long and stout anterior process is present, which expands distally but is not synostosed to the closely adjacent ectotympanic bone. From the base of the anterior process and ventral head emerges a substantial transversal lamina, which thickens ventrally towards the base of the manubrium, but there is no clear prominence here which could be regarded as an orbicular apophysis. The manubrium is oriented almost parallel to the anatomical axis, and it has no lateral process. Although there is no distinct muscular process, a large, fleshy tensor tympani could be seen in the scans inserting on the thickened part of the malleus near the base of the manubrium. The incus is relatively small with respect to the malleus (Table 2). Relative to the incus size, the short process is longer and thinner than in the other armadillos; its tip ends closer to the skull than in the other species, suggesting that the ligament anchoring it to the periotic is short and stubby. The long process is approximately perpendicular to the anatomical axis, and as in the other armadillos there is no lenticular apophysis. The stapes appears rotated relative to the long process (Fig. 6E). The stapes head is oval and continues into a neck that is shorter than those of the other species (Fig. 5E). Both crura are long, almost equally wide all along and inserted near the border of the footplate; they delimit a triangular intercrural foramen. The footplate is oval and relatively thick. The stapedial muscle is well‐developed and a large element of Paaw, almost as long as the footplate and nearly in contact with the stapes, was clearly visible (Fig. 4F).

Comparative morphometry

All the regression relationships between the middle ear measurements and the skull size of the species examined were found significant except for the stapedial measurements and the ML (Table 4). Although the slopes of the regression lines were positive, their values were below what would be expected from isometry in the cases of the stapes volume and stapes FA (Table 4), indicating that, in the larger species, those structures are relatively smaller than in the smaller species. Whereas points representing C. truncatus, Z. pichiy and Chaetophractus villosus were located on or close to the regression lines for all the parameters, Chaetophractus vellerosus and D. hybridus systematically fell above and below the line, respectively (Fig. 7).

Table 4.

Least‐squares linear regression relationships between middle ear parameters (y) and total skull length (x) in armadillos.

Parameter (y)	n	Intercept (a)	Slope (b)	p (Regression)	R 2	SE (b)	p (Allometry)
Middle ear cavity volume, mm3	11	−2.47	2.60	0.009	0.55	0.779	0.618
Malleus + incus volume, mm3	11	−4.47	2.77	<0.001	0.79	0.475	0.635
Stapes volume, mm3	9	−3.43	1.37	0.057	0.43	0.603	0.030
Tympanic membrane (pars tensa) area, mm2	11	−1.53	1.64	0.004	0.63	0.421	0.409
Stapes footplate area, mm2	11	−1.48	0.76	0.060	0.34	0.355	0.007
Malleus lever arm, mm	9	−0.62	0.65	0.055	0.43	0.284	0.261
Incus lever arm, mm	9	−0.89	0.61	0.026	0.538	0.217	0.118

Analyses were performed on log‐transformed data for both dependent and independent variables. The probability value of the slope coefficient b is indicated as p (Regression), and the coefficient of determination as R ²; SE(b) represents the standard error of b; the probability value obtained when testing the deviation of b from the theoretical value of isometric growth is indicated as p (Allometry). See text for further details.

Figure 7.

Relationships between middle ear parameters and skull size in armadillos, based on log‐transformed data. (A) Middle ear cavity volume vs. total skull length; (B) Volume of the malleo‐incudal complex vs. total skull length; (C) Stapes volume vs. total skull length; (D) Tympanic membrane pars tensa area vs. total skull length; (E) Stapes footplate area vs. total skull length; (F) Malleus lever arm vs. total skull length; (G) Incus lever arm vs. total skull length. The calculated regression lines are indicated as dotted lines. Members of the same species are ringed. See Table 4 for further information.

Area ratios varied considerably among species (Table 2). The lowest values were those of C. truncatus and D. hybridus, the highest were those of Chaetophractus villosus. Both pars tensa and stapes FAs were considerably larger in absolute terms in Chaetophractus vellerosus than in Z. pichiy (Table 2), although these armadillos have similarly‐sized skulls. The relatively long ML of Chaetophractus vellerosus gave it the highest LR (Table 2).

Discussion

Our findings regarding the morphology of the middle ear of C. truncatus and its relatives are broadly in agreement with previous descriptions in the literature (Doran, 1878; Fleischer, 1973; Segall, 1976; Novacek & Wyss, 1986; Patterson et al. 1989, 1992; Sidorkewicj & Casanave, 2012). We begin by comparing our results to these published descriptions.

Segall (1976) described a long, bony tube in Chlamyphorus which formed part of the external ear canal, composed of two successive segments joined by fibrous tissue. From his description, this tube would appear to be the result of ossification of the normally cartilaginous parts of the external auditory meatus. We did not find any such ossified tube extending from the recessus meatus in any of our Chlamyphorus skulls. Segall’s illustrations of the malleus show an ossicle completely lacking an anterior process, which he described as ‘short’, but our CT scans revealed the presence of a long anterior process, as noted by Fleischer (1973). Detachment of the malleus from the ectotympanic in armadillos usually involves the breakage of the anterior process (see also Sidorkewicj & Casanave, 2012; pers. obs.). This process, which is very narrow proximally in Chlamyphorus, had presumably snapped in Segall’s specimen. This probably also accounts for the relatively abbreviated processes illustrated for other armadillos by Patterson et al. (1992). We found a structure that could represent a markedly reduced transversal lamina only in one specimen of Chlamyphorus; Segall reported the presence of a small one, but this is largely a matter of interpretation of the nature of the structures at the junction of head, neck and anterior process.

The incus of Chlamyphorus was described and illustrated by Segall (1976) as having a finger‐like posterior crus (short process in the present paper), but our specimens had a conical short process. He did not mention the unusual nature of the incudo‐stapedial articulation, discussed later. He describes the stapes of Chlamyphorus as ‘of sauropsidan type, i.e. somewhat columellar’, perhaps referring to the long, flattened stapedial neck. Fleischer (1973) suggested that a secondary columellar morphology could arise in mammals from the crura inserting towards the middle of the footplate, rather than at its periphery. It is important, however, to distinguish the morphology of the armadillo stapedes, which have relatively wide intercrural foramina, from the imperforated or micro‐perforated stapedes found in sloths and the pygmy anteater Cyclopes (Doran, 1878; Novacek & Wyss, 1986; Patterson et al. 1992), which are more appropriately referred to as ‘columellar’. An imperforate stapes with very thin neck is also found in the adult naked mole‐rat Heterocephalus glaber (Mason et al. 2016).

Although some connective tissue strands were visible in the CT scans which might have represented vestiges, it seems that the tensor tympani muscle in C. truncatus is significantly reduced or absent. This has not previously been noted in this species. Although this represents an unusual feature among mammals in general, one of the middle ear muscles is commonly reduced or absent in subterranean species (Burda et al. 1992; Mason, 2001, 2013). The stapedius muscle was universally present in our armadillo specimens, and an element of Paaw was identified within it in at least one ear of our specimens of Dasypus and both Chaetophractus species, between muscle belly and tendon. This structure, first mentioned by Paaw (1615) in adult oxen, has frequently been mistaken for the lenticular apophysis of the incus (Graboyes et al. 2011). It has been described in species of bats, opossums, tree shrews, primates, carnivores, rodents and edentates (McClain, 1939; Henson, 1961; Hinchclifee & Pye, 1969; Pye, 1972; Wible, 2009). Among armadillos, it has been previously described in D. novemcinctus (Reinbach, 1952; Wible, 2010). Henson (1961) suggested that the element of Paaw might function to reduce friction between the stapedius tendon and the middle ear mucosa. This hypothesis is questionable, given that the element of Paaw is lacking in many mammalian groups. The intra‐specific variability that has been documented previously (Hinchclifee & Pye, 1969) and was also found in the present study suggests that the element of Paaw has no vital function.

Finally, Sidorkewicj & Casanave (2012) found a small pars flaccida in the Euphractinae. This membrane could not be seen directly in the prepared skulls examined. It seems likely that Dasypus has relatively the largest pars flaccida among the species studied, sealing the epitympanic recess lateral to the bodies of malleus and incus. In the other species, its area could be more precisely estimated from dorsal embayments in the bony tympanic rim. This notch was smaller in Chaetophractus villosus than in Chlamyphorus, insignificant in Zaedyus and absent in Chaetophractus vellerosus. The pars flaccida (Shrapnell’s membrane) is thought to be an ancestral feature among marsupial and placental mammals (Fleischer, 1978). Although it is present in some mammals known or suspected to have excellent low‐frequency hearing, including gerbils, jerboas and sengis (Lay, 1972; Mason, 2013), it is very small or absent in most subterranean mammals (Burda et al. 1992; Mason, 2006). The fossorial armadillo Chlamyphorus would therefore appear to be unusual in this respect.

Evolutionary and functional implications

Although Dasypus is second only to Chaetophractus villosus in size among the armadillos studied (Table 1), it has middle ear structures similar in size to those of C. truncatus, and in some cases smaller (Tables 3 and 2). Its middle ear shows a number of features likely to be ancestral for therians as a whole (Fleischer, 1978). These include the ectotympanic bone taking the form of an open ring which is only loosely connected to the skull by connective tissue. Novacek (1977) described for adult Dasypus the existence of a ‘cartilaginous cover of the ventral opening of the tympanic chamber’, which is rare among living mammals and was considered likely to represent a specialised condition. Small cartilaginous elements were found by other authors in prenatal specimens (e.g. van Kampen, 1915; van der Klaauw, 1922; Reinbach, 1952). In our prepared skulls, we did not find cartilage but there were several ossified entotympanic elements in this region, similar to what was reported by Patterson et al. (1989) and Wible (2010) in D. novemcinctus. The middle ear cavity volume is relatively small, both middle ear muscles are well‐developed and, as mentioned above, a large pars flaccida appears to be present. The malleus has a small head, large transversal lamina and a broad articulation with the ectotympanic by means of the anterior process; its manubrium is almost parallel to the anatomical axis. This conforms to Fleischer’s ‘ancestral’ ossicular morphology, found in species such as the opossum Didelphis and believed to be primitive for therian mammals. Patterson et al. (1992) found a very similar malleus morphology in the tolypeutines Priodontes, Cabassous and Tolypeutes. We agree with those authors that this is likely to have been the primitive ossicular morphology for armadillos.

The other armadillos examined here have complete bullae, the bulla being markedly extended into a bony external auditory meatus in the euphractines, and ossicles somewhere between ‘transitional’ and ‘freely mobile’ types. Those of Chaetophractus vellerosus come closest to the latter based on the prominent malleus head, small transversal lamina and the angle between manubrium and anatomical axis most nearly approaching 90°. These we interpret as derived characteristics for armadillos. Fleischer described freely mobile mallei as having abbreviated anterior processes, connected to the skull only through ligaments. This would contribute to a higher acoustic compliance, promoting low‐frequency sound transmission. However, in many mammals usually regarded as having ‘freely mobile’ ossicles, the anterior process actually retains a bony connection with the skull at its tip (Mason, 2006, 2016b; Mason et al. 2018). The bone of the anterior process tends to be very thin, and the articulation is flexible in fresh specimens. Armadillos represent a curious case in this respect: all retain very long anterior processes, the distal halves of which articulate with the ectotympanic bone. However, there is no synostosis except, in some species (C. truncatus, Chaetophractus vellerosus and Z. pichiy), at the very tip of the process. Whether the narrow, proximal part of the anterior process retains sufficient flexibility for armadillos to be regarded as functionally ‘freely mobile’ remains to be tested experimentally.

Billet et al. (2015) performed a detailed morphological study of the bony labyrinth of 17 extant xenarthran species, showing among other things that C. truncatus, Chaetophractus vellerosus, Z. pichiy and Euphractus sexcinctus are characterised by a large size of the fenestra vestibuli (oval window). This, they suggested, might be considered to be a synapomorphy supporting a placement of fairy armadillos close to euphractines. The oval window accommodates the stapes footplate, quite snugly in all armadillos in contrast to some subterranean rodents (Mason et al. 2010, 2016). Its relatively large size in Chlamyphorus and euphractines could be associated with their movement away from the ancestral ear morphology as found in Dasypus, which has a very small stapes footplate. This might have occurred convergently in the chlamyphorines and euphractines. Although the ossicles of both groups are similar, those of Chlamyphorus appear to be more flimsily constructed: the malleus head is small and excavated, the manubrium is thin and lacks a prominent lateral process and the stapes footplate has a very thin neck and a poorly‐developed labrum.

The malleo‐incudal articulation

Flattened malleo‐incudal articulations have been reported in many fossorial mammals (Segall, 1973; Burda et al. 1992). This may be linked to the fact that variations in static air pressure in the external ear canal cause translational (inward‐outward) movements of the malleus (Hüttenbrink, 1988): the flattened articulation in subterranean mammals might help their ossicles to accommodate to air pressure changes experienced underground (Mason, 2006). Segall (1976) reported that C. truncatus has a malleo‐incudal joint intermediate between those of more exclusively subterranean and non‐fossorial species. We found it to be more flattened than in the other armadillos studied here.

The incudo‐stapedial articulation

In most mammals, the incudo‐stapedial articulation involves a lenticular apophysis: a disc‐like, bony process, connected to the end of the long process of the incus by means of a thin pedicle. Its medial face represents the facet for articulation with the stapes. Surprisingly, we found no pedicellate lenticular apophysis in any armadillo specimen. It was clear from our scans and reconstructions that the stapes instead articulates directly with the flattened, spatulate end of the incudal long process. We suspect that previous descriptions of a lenticular apophysis in armadillos (Fleischer, 1973; Sidorkewicj & Casanave, 2012) were based either on interpreting the flattened articulation facet on the long process of the incus as the apophysis, or on specimens in which the stapes head had broken from the stapes body and remained in articulation with the incus. The lack of a pedicellate lenticular apophysis is highly unusual among mammals but has previously been documented in monotremes, cetaceans and sirenians (Fleischer, 1973, 1978).

The thin pedicle connecting the lenticular apophysis to the long process of the incus in most mammals is predicted to confer significant flexibility (Funnell et al. 2005). Especially long pedicles have been documented in the saltatorial rodent Jaculus (Mason, 2016b) and in the subterranean mole‐rat Spalax, which communicates with conspecifics by head‐thumping on its burrow walls (Mason et al. 2010). A long pedicle might help to decouple the stapes and hence the inner ear from impacts affecting the malleus and incus. A similar function in armadillos might be served by the very thin neck of the stapes, which is made of essentially laminar bone and is likely to be quite flexible when hydrated.

Middle ear cavity volume

Middle ear cavity volumes can vary greatly among mammals, even within the same family or in related groups (Lay, 1972; Webster & Webster, 1975; Mason, 2013, 2016b). Enlarged middle ear cavity volumes are expected to benefit low‐frequency hearing in small mammals (reviewed by Mason, 2016b). This is because cavity compliance, which is proportional to volume, tends to dominate overall middle ear compliance in small mammals, and this limits sound transmission at low frequencies (Ravicz et al. 1992). In some subterranean species, enlarged middle ear cavities could represent an adaptation for underground vocal communication (Schleich & Vassallo, 2003). However, middle ear cavity volume is not particularly large in subterranean mammals (Mason, 2001): the largest middle ear cavity volumes relative to body size are actually found in species from arid regions, including gerbils, kangaroo rats, chinchillas and certain sengis (Lay, 1972; Webster & Webster, 1975; Mason, 2013, 2016b). This may be advantageous given that lower frequencies propagate better than higher ones in arid environments (Huang et al. 2002; Rosowski et al. 2006).

Among armadillos studied, the species with by far the largest middle ear cavity volume in both absolute and relative terms was Chaetophractus vellerosus (Table 3; Fig. 7A). Its capacious middle ear is based in part on the great dorsal expansion of the epitympanic recess, as well as the presence of a mastoid cavity. The tympanic cavity proper is also large (note that absolute values are similar to those of Chaetophractus villosus; Table 3).

Petter (1953) measured bullar length in gerbils and presented this as a percentage of skull length, in order to compare relative sizes between species. In their studies on armadillos, Roig (1972) and Squarcia et al. (2007) referred to this percentage value as the ‘bullar hypertrophy index’ (BHI). Roig established three well‐defined groups based on BHI, and related the degree of hypertrophy of the species with the aridity of the environment they inhabit. Groups were: (i) species without hypertrophied bullae, which live in relatively damp environments; (ii) species with moderately hypertrophied bullae, which range in their distribution from semi‐humid to semi‐arid environments; and (iii) species with more hypertrophied bullae, which are typically inhabitants of semi‐arid and arid environments. Roig placed D. hybridus and D. novemcinctus (mean BHI: 6.42% in both cases) within Group 1, together with Priodontes giganteus (= P. maximus; mean BHI: 4.80%), a typical inhabitant of warm and humid regions, and a species referred to as ‘Cabassous loricatus’ (mean BHI: 9.09%), the identity of which is unclear. Group 2 included only Tolypeutes matacus and Euphractus sexcinctus (mean BHI: 10.47% and 12.59%, respectively), whereas Group 3 encompassed Chaetophractus, Zaedyus and Chlamyphorus species. The largest BHI found by Roig was that of Chaetophractus vellerosus (17.28%). Similar results were reported by Squarcia et al. (2007) for Chaetophractus and Zaedyus species. Our reconstructions allowed us to measure the actual middle ear cavity volumes, which are affected not just by external bullar dimensions but also by expansions of the cavities into the surrounding bones of the skull. We found that, among the species we studied, Chaetophractus vellerosus had the largest middle ear cavity volume while D. hybridus had the smallest, relative to skull size (Fig. 7).

Although some degree of overlap in the distribution of the species occurs within central Argentina, D. hybridus appears to be more influenced by levels of precipitation than are the other armadillos considered here. It is the species with highest probability of occurrence in north‐eastern areas of the country (Abba et al. 2012), where the mean annual precipitation exceeds 1400 mm (based on climatological data from Bianchi & Cravero, 2017). Chaetophractus vellerosus, on the other hand, is a typical inhabitant of xeric habitats from low to high elevations (Wetzel et al. 2007), with high probability of occurrence in the north‐west of Argentina (Abba et al. 2012). Middle ear cavity expansion in Chaetophractus vellerosus may therefore be related to the aridity of its habitat, echoing Roig’s (1972) conclusion. Based on its expanded middle ear cavities, we predict that Chaetophractus vellerosus has the best low‐frequency hearing among the species examined, but it would be premature to link this with the loud distress calls from which it derives its English name, screaming hairy armadillo (Amaya et al. 2019). Within the group Chaetophractus villosus ‐‐ Z. pichiy ‐‐ C. truncatus, although the correlation between middle ear cavity volume and aridity of habitat is less clear, their relatively large middle ear cavities and ossicular structures suggest some degree of adaptation to low‐frequency hearing, relative to the ancestral state found in Dasypus. The retention of ancestral ear structures in Dasypus might represent phylogenetic inertia rather than an adaptive fit to a specific type of environment.

Area and lever ratios

Anatomical area and LRs have been classically used in ‘ideal transformer’ models of the impedance‐matching function of the mammalian middle ear (e.g. Dallos, 1973). Mason (2001) found that fossorial mammals tend to have lower area ratios than non‐fossorials, as a result of relatively large stapes FAs. LRs also tend to be lower in fossorials although MLs and ILs considered individually do not differ significantly in length between fossorial and non‐fossorial mammals (Mason, 2001). The mean area and LRs found here in C. truncatus (21.79 and 2.19, respectively) were intermediate between the mean values reported by Mason for fossorial and non‐fossorial mammals (area ratios: 17.11 and 28.27, respectively; LRs: 1.72 and 2.24, respectively). Although the anatomical area ratio in Chlamyphorus was the lowest among the armadillos studied, its LR was only exceeded by those of both Chaetophractus species.

‘Ideal transformer’ models of middle ear function are subject to significant criticism based on their oversimplifications, and their predictions are often not supported by experimental measurements (see Mason, 2016a for a recent review). Middle ear function is greatly complicated by the flexibility and frequency‐dependent vibrational characteristics of the tympanic membrane and ossicular chain. Therefore, although gross differences in anatomical area and LRs among mammals can probably still tell us something about the nature of their hearing, small differences such as those found among the armadillos in this study are of doubtful functional significance.

Conclusion

In comparison with the middle ear apparatus of Dasypus, which is interpreted here as being primitive for Cingulata, the pink fairy armadillo C. truncatus shows some derived characteristics. These include the development of a complete auditory bulla which accommodates a relatively larger middle ear cavity volume, a reduced pars flaccida, ossicles which have moved towards a ‘freely mobile’ morphology, the reduction or loss of the tensor tympani muscle and a more flattened malleo‐incudal articulation. This mirrors the direction of evolution documented among the subterranean talpid moles, in comparison with their more terrestrial relatives (Mason, 2006). However, the first three of these characteristics are shared with euphractine armadillos and so cannot necessarily be considered to represent adaptations to the more exclusively subterranean environment of Chlamyphorus. All of these armadillos including Chlamyphorus have long anterior processes and lack pedicellate lenticular apophyses, neither of which is expected of subterranean mammals. The reduction of the tensor tympani and the thin, flimsy appearance of the ossicles of Chlamyphorus might in fact be indicative of degeneration of the ear: in this respect there are interesting parallels with the naked mole‐rat, Heterocephalus glaber, the small, delicate ossicles of which have also been considered potentially degenerate (Mason et al. 2016). In terms of low‐frequency specialisation, it is actually Chaetophractus vellerosus which stands out among armadillos, based on its voluminous middle ear cavity. Given its association with xeric habitats, this can be added to the list of arid‐region mammals with markedly hypertrophied middle ear cavities.

Conflict of interest

None declared.

Supporting information

Appendix S1 Details of the scans made in the present study.