Syrinx
Figure 1
Palaeognath syringeal elements with tracheosyringeal cartilages differentiated by brown shading (B-E), interannular tissues are shaded in grey. A-B, Cassowary, Casuarius casuarius, FUR180; ventral, scale bar = 10mm. C, Tinamou, Nothura darwinii, adapted from Garitano-Zavala [39, fig. 1C]; ventral. D, Rhea, Rhea americana, adapted from Forbes [18, fig. 7, 8]; ventral and dorsal. E, Kiwi, Apteryx mantelli, adapted from Forbes [18, fig. 3], ventral. F, Ostrich, Struthio camelus, adapted from Forbes [18, fig. 1]; ventral. Abbreviations: bs.cs: bronchosyringeal cartilages, ia.i: interannular interval, tr.c: tracheal cartilages, trs.cs: tracheosyringeal cartilages. C-F scaled to same size approximately.
The cartilaginous syrinx of our female cassowary (FUR180) structurally conforms to the tracheo-bronchial syrinx type (figure 1A-B); as do all palaeognaths [20 p. 61]. The syrinx of FUR180 is simple and comparable in form with others previously described, such as the adult male assessed by King [19 p. 124], indicating the absence of sexual dimorphism in this organ. Of the palaeognaths, and all birds, the ostrich has been considered to have one of the simplest syrinxes, with which it can produce a limited repertoire of sounds (figure 1F) [40]. However, the ostrich syrinx does have a pessuliform process and potentially a tympanum [18, 41], indicating a more derived state than in other palaeognaths including the cassowary. The syrinx of the rhea (figure 1D) is the most complex, due to the presence of specialised anatomical structures, including fully developed intrinsic musculature, absent in most other palaeognaths [18, 21, 42].
Cart. tracheosyringeales- There are five tracheosyringeal cartilages associated with the cassowary syrinx, distinct from preceding true tracheal cartilages as they are thinner ventrally and laterally. Dorsally all are incomplete along the midline, with the extremities bending medially towards the centre of the syrinx, a result of tracheosyringeal membranes contracting post-death [21]. Forbes [18] and Pycraft [21] noted the presence of these imperfect tracheosyringeal cartilages, proposing the cranio-caudal space formed between the cartilage extremities is occupied by transversely running fibrous and elastic tissues [21] later termed the tracheosyringeal membranes (mem. tracheosyringealis) [19 p. 128]. Dorsally incomplete cartilages are present in the ostrich, moa, kiwi and emu, although lacking in the rhea and tinamou. The number of incomplete cartilages varies depending on the species, although no other species approaches having nine, the number present in the cassowary specimen. Three have been noted for the moa [43] and emu [18, 21], and two for the ostrich [18]. The number present in the kiwi is dependent on the species; Apteryx australis has three compared to the single incomplete cartilage in A. mantelli [18].
All five tracheosyringeal cartilages angle caudally along the medial line of the ventral side of the cassowary syrinx, with the degree of the angle increasing caudally with each cartilage, paired with a cranio-caudal increase of width. These features are common in palaeognaths, although variable in some taxa such as the ostrich and kiwi, which develop this character on the dorsal side of the syrinx. Ventral modification to the most caudal tracheosyringeal cartilages in the moa differentiate this taxon from others: the cartilage lengthens cranio-caudally at the most caudal point of the V, from which a caudo-medially directed projection extends [43]. Oliver [43] interpreted this keel syringeal ring to likely be the ventral attachment point of a pessulus.
Noted as a common feature among casuariids (Dromaius and Casuarius) by Pycraft [21], the cassowary syrinx shows poor transitional definition between bronchosyringeal and tracheosyringeal cartilages. The most caudal tracheosyringeal cartilage (trs. cs. 1) closely reflects the structure of the first bronchosyringeal cartilage with cartilages differentiated only by partial fusion of the ventral extremities present in the former, maintaining a single element structure. Tinamous also display a gradual transition between cartilage types [39], although greater transitional definition is noted in the ostrich, kiwi and rhea. The ostrich trachea increases in diameter in the few cartilages preceding tracheal bifurcation; the following bronchosyringeal cartilages are much narrower craniocaudally [18, 41]. Alternatively, distinction between the two cartilage types in kiwi is formed from a widening of the bronchosyringeal cartilages after tracheal bifurcation [18, 21]. The transitional definition in the rhea is unique among palaeognaths with the fusion of tracheosyringeal cartilages forming a tympanum just cranial to tracheal bifurcation [18, 21, 42].
Pessulus- No pessulus is present in the cassowary specimen FUR180, with the left and right medial tympaniform membranes fusing transversely along the dorso-ventral plane, at the level of tracheal bifurcation. As expected, no intrinsic musculature was found in the µCT-generated model as expected given its absence in specimens described by Forbes [18] and Pycraft [21]. Similarly, the tinamous [39], kiwi [18, 21] and emu [18] also lack a pessulus. In the rhea, this structure is present; the pessulus links the caudo-medial point of the dorsal and ventral sides of the tympanum as a narrow osseous bridge [18]. Ossification has been recorded for rhea, despite the structure being primarily cartilaginous, suggesting that increased ossification occurs later in ontogenetic staged in males. Although we found no support for this in the NMNZ collection, based on a described structure by Owen [44], Oliver [43] reported that moa also develop an ossified pessulus forming a partial bridge across the ring and likely completed by cartilage. Early observations of the ostrich syrinx found the third tracheosyringeal cartilage to contain a short caudal projection, medially on the ventral border [18]. This is considered a pessuliform process; not a true pessulus due to it not traversing the ventrodorsal width of the trachea. Yildiz and colleagues [41] noted the presence of a double-folded structure formed of connective tissue, suggesting this may act similarly to a true pessulus, providing support for the medial tympaniform membranes.
Tympanum- FUR180 has complete lack of fusion between tracheosyringeal cartilages indicating the tympanum is absent in cassowaries. No evidence was found to support Forbes’ [18] claim for the presence of an ‘expanded’ tympanum, and it is unlikely that incomplete and unfused cartilages such as we observe could function similarly to a true tympanum.
Of all palaeognath taxa, a tympanum has been described in the ostrich, rhea, and moa; this structure is absent in tinamous [39]. However, among these taxa, only in rhea has the presence been confirmed with complete dorsal and ventral fusion of four to six tracheosyringeal cartilages [18, 21, 42]. Fusion between tracheosyringeal elements has been described in some ostrich specimens, with the tympanum comprising three tracheal cartilages, although Yildiz et al. [41] found the cartilages only appeared to be fused through the presence of ligamentum annulare. Oliver [43] also provides a description pertaining to the presence of a possible tympanum in moas, located cranial to the tracheosyringeal cartilages [45 p. 107]. We searched numerous moa specimens and provide in the supplementary data (SI.1) a list by species and presence of tracheal rings, noting the type of rings present including the unique syringeal keeled ring that represents an incomplete ossified pessulus. The tympanum as described by Oliver [43, fig 26] was not found among any tracheal ring sets present in any moa taxa (SI. 1.). We consider that given the keeled syringeal ring is invariably present, then there was no tympanum in moa, given such would incorporate this ring with more than one other, making an even more robust ossified element.
Cart. bronchosyringeales- The cassowary bronchi are asymmetrical, with the left bronchus larger in diameter. Also contributing the asymmetry, the ventral extremity of the third bronchosyringeal cartilage on the left bronchi is angled ventro-medially and expands caudally, both characters are absent on the right bronchi, although both sides display an increase in cartilage length from preceding cartilages. The medial extremities of the bronchosyringeal cartilages overlap dorsally and ventrally, indicating a potential for expansion of interannular intervals during use. Interannular intervals (spaces) between bronchosyringeal cartilages remain relatively uniform for the length of the specimen, with the absence of large, paired intervals indicating lateral tympaniform membranes do not develop. The kiwi similarly possesses uniform intervals [18], whereas in the ostrich, ventral intervals narrow, and cranial intervals widen [18, 41]. The tinamou possesses two wide interannular intervals between the first to the fourth bronchosyringeal cartilages [39]. The first two bronchosyringeal intervals indicate the presence of the lateral tympaniform membrane, also noted for the rhea.
Hyoid
Figure 2
Palaeognath hyoid elements. A-B, Cassowary, Casuarius casuarius, FUR180; dorsal view, A- scale bar = 10mm. C, Tinamou, Nothoprocta perdicaria, NMNZ S. 22983; dorsal view, scale bar = 10mm. D, Moa, Megalapteryx, NMNZ S. 400; dorso-lateral view, scale bar = 10mm. E, Emu, Dromaius novaehollandiae, adapted from Parker [36, plate XII]; dorsal view. F, Rhea, Rhea americana, adapted from Parker [36, plate X]; dorsal view. G, Ostrich, Struthio camelus, adapted from Soley et al. [22, fig. 3]; dorsal view. Abbreviations: bh: basihyal, buh, basiurohyal, cb: ceratobranchial, epb, epibranchial, ur: urohyal.
All typical hyoid skeletal components are present within the hyoid apparatus of FUR180 (figure 2A-B). The basihyal, urohyal, and a paraglossal are cartilaginous, with only the ceratobranchials ossified. Ossification of ceratobranchials is common to all Aves, while the extent to which other elements ossify varies by lineage [22, 24 p. 367, 45 p. 110, 46, 47].
Basiurohyal- The joint between the basihyal and urohyal is indiscernible, with complete fusion of the two cartilaginous skeletal elements forming the basiurohyal (figure 2B). This character is common among palaeognaths; the only exception is the rhea where the urohyal is lost (figure 2F) [24 p. 369, 36, 48]. Rostrally, the basiurohyal curves dorsally, preceding a slight ventral arch centrally along the corpus, with the caudal point terminating with a minor inwards hook towards the laryngeal cricoid cartilage. The basiurohyal in the cassowary and emu have rounded tips [27], compared to that of the ostrich (figure 2G) which tapers caudally to a pointed tip, and rostrally terminates in two bulbous projections divided by a shallow notch [22, 27, 36]. The basihyal of the rhea is more cylindrical than in other palaeognaths although it also terminates in a rounded rostral tip [24 p. 366, 48].
Despite the basiurohyal being completely unossified in the cassowary, ossification of basiurohyal elements does occur in the kiwi [21] and rhea [48]. A partially ossified basiurohyal was also identified in a single moa specimen (Megalapteryx, specimen no. NMNZ S.400). The identification of an ossified moa basiurohyal shows that partial ossification may exist in at least the basal moa genus Megalapteryx though a lack of these elements in the moa fossil record limit our ability to determine whether this is an incident cause by the age of the specimen. We also identified a basiurohyal in which the urohyal and caudal portion of the basihyal were ossified in a tinamou (figure 2C) (Nothoprocta perthicaria, NMNZ S.22983). In an earlier study, Li et al. [47] found that despite midline ossification being a key component of the hyoid apparatus in neognath birds, the degree of ossification in palaeognaths varies and is often incomplete when present. This is correct for most palaeognaths, however even our limited observation of tinamou and moa specimens show complete ossification can be present. Therefore, further investigations across a wider sample of taxa is required to fully assess this variability.
Ceratobranchiale and epibranchiale- The basiurohyal articulates mid-way along its lateral edges with the ceratobranchials. The bulbous, disk-like proximal ends of the ceratobranchials sit low in the concave basiurohyal sockets; the articular surface of the sockets is larger than that of the proximal ceratobranchial end, indicating an allowance for considerable movement within the joint. The ceratobranchials are elongate with one third of their length extending past the caudal point of the basiurohyal. The shaft curves dorsally towards the ceratobranchial-epibranchial joint, from which the shorter cartilaginous epibranchials extend caudo-dorsally. In our specimen the right epibranchial is deformed with a large, sharp medial bend and increased tissue mass, contrasting with the smooth curve of the left epibranchial. The lack of symmetry, as well as no previous mention of such deformity within the literature for the cassowary or any other palaeognath taxa, indicates a pathology. The shape would impede the functionality of the epibranchial, including movement in and out of the hyoid sheath.
Palaeognath ceratobranchials are often cylindrical in shape, although can be slightly flattened as in the rhea [36]. Ostrich epibranchials are elongate [24 p. 371] compared to most other palaeognaths, which have short epibranchials relative to the ceratobranchials [24 p. 367]. In the tinamou, both epibranchials and ceratobranchials display increased elongation [36] compared to other taxa. We found the tinamou to have ossified epibranchials, differing from other palaeognaths including its closest relative, the moa, in which the epibranchials do not ossify (figure 2D).
Paraglossum- The cranial portion of the hyoid skeleton attaches to the paraglossal; both structures are encased by soft tissue, connecting the paraglossal to the back of the tongue body. The cassowary paraglossal is a single un-ossified element seemingly with a rounded triangular shape, as suggested by Parker [36]. The paraglossal of the emu (figure 2E) is similarly tear-drop shaped, although the caudal edge may be rounded or scalloped [27, 36, 49]. The shape of the paraglossal in the rhea reflects the more triangular shape of the tongue, although it is smaller and with an oval opening dorsal on the palate [48]. The tinamou paraglossal is much narrower in width than other species, with scalloped margins and two caudally directed projections, one from each caudo-lateral corner [24 p. 372, 36]. Unique among palaeognaths, the ostrich paraglossal is divided into two narrow, caudo-laterally directed individual paraglossia, situated ventro-laterally in the tongue body [24 p. 371, 27, 50]. As the ostrich is phylogenetically basal among palaeognaths, this paired state could potentially be plesiomorphic, the rhea then shows partial fusion and the other palaeognaths, complete fusion.
Lingual corpus- When compared to the tongues of other avian taxa, palaeognath tongues are significantly shorter relative to the mandible; they have thus been described as vestigial organs, rudimentary in morphology [28, 50]. The cassowary tongue is no exception, reflecting the limited role played by the tongue during the ‘catch and throw’ feeding method, a method utilising obligate inertial feeding in which the tongue is unrequired [23 pp. 53 and 74, 51, 52].
As in the rhea, emu, and ostrich (24 pp. 366, 370 and 371, 28, 48, 49, 53], the cassowary tongue is cranio-caudally flattened and triangular. Only the tongue of the kiwi varies significantly, being elongate [54], reflecting the shape of the long and narrow bill. The kiwi elongate tongue is likely a result of dietary specialisation, with the kiwi diet consisting primarily of invertebrates [55]; the elongated bill is required for detecting buried or submerged prey using vibration-sensitive mechanoreceptors [56]. The cassowary tongue corpus has a smooth and rounded rostral apex as in tinamous and ostriches, although varying from the pointed tip of the rhea tongue [24 p. 372, 48]. The caudal and rostral edges of the tongue are concave, although the rostral notch is more prevalent in the rhea, ostrich, and tinamou, than the cassowary and emu [24 p. 372, 48, 57].
Numerous lingual papillae, arranged asymmetrically, line the sides of the cassowary tongue, increasing in length and width caudally towards the tongue base. Although the emu displays analogous structures, these are not present in all palaeognaths. The tongue of the rhea lacks lateral papillae although the caudo-lateral corners project caudally [28, 49, 48, 57]. Both absence and presence of papillae have been noted for the ostrich [24 p. 371, 27], and papillae are completely absent in tinamous [24 p. 372]. Only the emu tongue is considered to have caudal papillae, although poorly defined and rudimentary when compared to those directed laterally [29, 49]. Lingual papillae are absent on the dorsal surfaces of all palaeognathous tongues [28].
Larynx
Figure 3
Palaeognath laryngeal elements. A-B, Cassowary, Casuarius casuarius, FUR180; dorsal view, scale bar = 10mm. C, Rhea, Rhea americana, adapted from Crole and Soley [48, fig. 7]; dorsal view. D, Ostrich, Struthio camelus, adapted from Tadjalli [46, fig. 9a, b]; dorsal view. E, Tinamou, Nothura maculosa, MMC321; ventral view. F, Moa, Euryapteryx curtus, NMNZ S. 44757; lateral and ventral views. G, Kiwi, Apteryx rowi, NMNZ OR.27243A; dorsal view. Abbreviations: ar: arytenoids, cr.c: cricoid cartilage, cr.w: cricoid wings, oss: ossification, pc: procricoid, tr.c: tracheal cartilages.
The larynx of the cassowary FUR180 (figure 3A-B) has a standard avian anatomy, composed of the cartilaginous skeletal elements, the cricoid, procricoid, and paired arytenoid cartilages. FUR180 is entirely cartilaginous as in the emu, some palaeognaths (ostrich [46], rhea [48] and kiwi) have poorly and variably defined ossification centres in the corpus, with partial ossification in the rhea [48] and kiwi (personal observation, figure 3) likely dependent on ontogenetic stage. Only tinamous and moa have a strongly ossified cricoid corpus where the entire corpus is well ossified and has well defined margins resulting in a distinctive cricoid bone. This character is thus identified as a synapomorphy supporting the molecular-based pairing of this clade. To date, morphological characters supporting the molecular identification of this clade [3, 7, 38] have remained elusive and thus, these findings are significant and provide phylogenetically informative data.
Cricoid cartilage- The cassowary cricoid FUR180 is characterised by a concave plate- or basin-like corpus, with a smooth dorsal and ventral surface. Neither the cassowary, emu, rhea, nor ostrich develop a median ridge traversing the dorsal (inner) surface of the main ventral plate/bowl of the cricoid [26 p. 73, 46, 48]. This crista is present in the kiwi [26 p. 72], moa and tinamous, projecting dorsally into the laryngeal lumen. Some moa species have two ridges of varying heights [58, 45 p. 106], similar to the cricoid of N. maculosa which has two slightly-raised ridges. The inverse of these ridges are recognisable on the ventral surface of the cricoid in both taxa (Figure 3E, F). The lateral margins of the Moa are smooth (figure 3F); this is also noted within the tinamou species E. elegans, although varying from the scalloped margins of the cricoid in N. maculosa (figure 3E). A small medially located cartilaginous caudal projection has been observed in the rhea (figure 3C). This projection is often fused with tracheal cartilages and, in some species, is situated between two smaller, caudo-medially directed extensions [48]. Many moa genera, including Dinornis and Pachyornis, also develop these features [43, 58]. Caudal projections are absent in the cassowary. The emu and ostrich develop a rostral process [22, 26 p. 73, 46] which ossifies in the ostrich, acting as an attachment point for the cartilaginous basiurohyal.
Cartilaginous cricoid ‘wings’ extend seamlessly dorsocaudally from the lateral margins of the cricoid cartilage in the cassowary, ostrich, emu, and rhea. As no moa cricoid wings have been identified, despite the collection of multiple cricoid bones from fossil deposits, it is likely this element was also cartilaginous in moa. Ossification of the cricoid in moa [43, 45 p. 106] and tinamou (observations herein), indicates that the wings fused to the lateral borders of the cricoid were cartilagenous. This hypothesis is supported by images of a tinamou cricoid (Marcos Cenizo, Museo de Historia Natural de La Pampa). In all palaeognath taxa, the cricoid wings narrow dorsally and are directed caudally [26 p. 73, 46, 48]. In the tinamou, rhea [48], emu and ostrich [22, 26 p. 73, 46], the cricoid wings join dorsally, completing the cricoid ring caudal to the procricoid. In the cassowary, the wings do not articulate dorsally; instead the procricoid and a cranial projection from the medial point on the dorsal side of the second tracheal cartilage insert between the wing extremities.
Cart. procricoidea- The cassowary procricoid cartilage is formed of a flattened rectangular corpus with a distal, cranially extending head, and a proximal, caudally directed tail. The tail is triangular, with the cricoid-procricoid joints on the flattened dorsal edge. The head is rounded cranially and flattened laterally, forming the dorso-medial walls of the concave basins which run ventrally along both sides of the procricoid. This concavity supports the arytenoids which extend rostro-laterally from the procricoid. Both the procricoid and paired crico-procricoid joints are seemingly caudo-medially supported by the cranial extension of the second tracheal cartilage. Dorsally, the emu procricoid is a simple rectangular shape [26 p. 73], whereas the shape is wide and rounded rostrally in the ostrich. The ostrich procricoid also develops a ventro-caudally directed projection which extends between the cricoid wings [22]. The rhea procricoid is similar in shape although more angled, with a flattened rostral margin, and dorsally triangular caudal projection [48]. The dorso-cranial procricoid head, seen in the cassowary, is possibly absent in other palaeognaths, as is the cranial projection from the tracheal cartilages which sits below the procricoid. However, the literature provides no information on the procricoid for the tinamou, elephant bird or moa, with the descriptions for ostrich, rhea, and emu procricoids, brief and lacking detail.
Cart. arytenoidea- In the cassowary, the arytenoid corpus is cartilaginous, flat and elongated, extending laterally and ventrally to form two sides of a V-shape. The caudal end faces medially towards the opposing arytenoid and rests within the lateral procricoid joint concavities. Through the corpus and rostral projection, the flattened sides twist laterally to face dorsally. This arytenoid structure is shared with the emu, the closest relative of the cassowary although they vary with the lateral margins of the emu arytenoid converging rostrally into a cranio-medial point [26 p. 73]. The description of the arytenoid cartilage in the moa [43] suggests a similar structure, varying primarily in that the moa arytenoids are partly ossified. However, in the few preserved moa arytenoids in the NMNZ show more extreme curvature throughout the bone. In contrast, the arytenoids of the rhea are formed from elongated, paired bars with proximally directed projections extending from the caudo-dorsal aspect of the arytenoids for attachment to the procricoid [48]. The ostrich arytenoids are also formed of elongate, paired bars, although have thin cartilaginous plates extending from the lateral margins, unique among palaeognaths. The plates form two lateral, and one dorsal projection with smooth, rounded margins.
Glottis- The arytenoids are covered in a dense mucosa, which forms the glottis mound. The dorsal surface of the glottis mound is typically smooth in palaeognaths including the cassowary; the only exception being the tinamous [26 p. 70]. Prominent laryngeal papillae extend from a widened caudal margin of the glottis mound in the rhea [48, 57, and kiwi [54]; the shape of the caudal papillae vary with angular, rounded, and rectangular forms (54). The lateral and dorsal projections of the arytenoid in the ostrich support the mucosal embellishments of the glottis mound, forming what has been termed a star-shape [22, 27, 49]. The lips of the glottis in both the ostrich and the rhea are supported internally by the arytenoids [22, 27, 48], however the glottis lips of the cassowary and emu [27] are not supported by the arytenoids but instead are formed of a separate mucosal structure, with a layer of dense musculature between.
Morphological Character Optimisation
To assess the phylogenetic signal of syringeal, hyoidal, and laryngeal (SHL) characters, we compared their fit to morphology-only, molecular-only, and combined data trees (see methods for full details). A fair comparison of character fit across trees is difficult due to radically different taxon sampling: notably most fossil taxa are missing from the molecular-only trees. However, the most important difference in the topologies concern relationships between major clades of ratites. Hence, we used the topology of a combined morphological and molecular data analysis, newly performed here, as one tree for comparison; to generate the other trees, we then re-arranged the major palaeognath groups to conform to either the morphology-only tree [59], or the molecular-only tree [3]. Lithornis was not sampled for the molecular tree and so thus left in its basal position as per the combined tree; however, this taxon is not codable for most SHL characters and so has very little impact on results (see supplementary data for results of phylogenetic analyses and character optimisation). For each topology, overall fit, as well as apomorphic and homoplasious states were identified. The results displayed in table 1 show the syrinx, hyoid and larynx characters to have a higher affinity for the combined-data topology than either the molecular- or morphological-only topologies. The tree length is lower for the combined-data topology (102), with results also showing a higher consistency index (CI= 0.5196) and lower character homoplasy (HI= 0.4804), both leading to a higher retention index (RI= 0.6818), the proportion of taxa with non-homoplasious states. The number of unique and unreversed apomorphic characters (CI = 1.0) is no less than 10 for the three topologies, although the data again favours the phylogenetic relationships obtained from the combined-data (AC = 14). Optimisation results for the morphological topology support the SHL data to have the lowest affinity for the morphological phylogenetic relationships.
Table 1:
Optimisation of SHL palaeognath characters on morphological, molecular and combined-data topologies. Tree index statistics for each optimised topology, including CI (consistency index), HI (homoplasy index), and RI (retention index), as well as AC (unique and unreversed apomorphic characters, CI = 1.0).
Topology
|
Tree Length
|
CI
|
HI
|
RI
|
AC
|
Morphological
|
108
|
0.4907
|
0.5093
|
0.6429
|
10
|
Molecular
|
106
|
0.5000
|
0.5000
|
0.6558
|
11
|
Combined Data
(Mor + Mol)
|
102
|
0.5196
|
0.4804
|
0.6818
|
14
|
Figure 4:
Syrinx, hyoid, and larynx character optimisation onto morphology-only, molecular-only, and combined-data topologies. Characters identified include those optimised as unique and unreversed, unambiguous characters (black filled circles), homoplasious unambiguous characters (empty black circles), and unique and unreversed, ambiguous characters (grey filled circles). A, Morphological topology optimisation. B, Molecular topology optimisation. C, Combined-data topology optimisation.
In the following discussion, ambiguous changes are those which are optimisation-dependent (e.g. vary across deltran or acctran), and unique and unreversed characters are those with a CI of 1. We discuss and present (Figure 4) deltran results, but flag the optimisation-dependent changes as ambiguous. The three topologies show similar results for (homoplasious) autapomorphic character changes defining cassowaries. Three autapomorphic character changes are identified for all topologies: character 5, 2 --> 0 (bronchosyringeal cartilages wider at medial ends); character 6, 0 --> 1 (minor asymmetry present); character 11, 0 --> 1 (caudal end of the trachea almost cylindrical). All these changes have ambiguous optimisation due to missing data from Casuarius bennetti, (these changes might define the C. bennetti plus C. casuarius clade or C. casuarius alone). In all three phylogenies, 5 character changes consistently united both the cassowary and emu lineages: character 17, 1 --> 0 (cartilaginous basiurohyal); character 26, 1 ==> 2 (Numerous lingual papillae along the lateral margins of the tongue corpus); character 37, 1 --> 2 (cartilaginous arytenoids); character 40, 0 --> 1 (Arytenoid cartilage a separate structure to the glottis lips); character 41, 1 ==> 0 (No laryngeal papillae present on the glottis lips). The changes for characters 26 and 40 are unique and unreversed for all three topologies.
As a result of missing data for the elephant bird syrinx, hyoid and larynx, there are no characters supporting the kiwi and elephant bird clade retrieved in the combined-data and molecular data topologies. Additionally, the character states identified as apomorphic for the kiwi are optimisation ambiguous as they might apply to the kiwi/elephant bird clade. In all three topologies, one of these characters is unique and unreversed for the kiwi: character 24, state change 0 to 1 (reduced ovular, tongue shape, dorsal view).
The tinamou-moa clade, robustly supported by DNA data but not predicted by traditional (skeletal) morphological trait, has new support from the SHL traits. This includes the ossification the cricoid cartilage (character 29, state 2) and the arytenoids (character 37, state 0). The character 37 change is an unambiguous optimised synapomorphy in both the molecular and combined-data topologies; this change is not unique and unreversed however as while unique within palaeognaths, it occurs outside the group, in sampled neognaths (Grus, Gallus, Anseranas). The state change for character 29 is optimised similarly for the molecular topology, however due to the basal placement of the tinamou and moa in the combined-data topology, a lack of complete cricoid ossification is optimised as an unambiguous synapomorphy for the crown palaeognaths, excluding the tinamou/moa clade. The functional significance of these ossifications is not obvious. Partial ossification of the cricoid has been recognised in older birds in taxa such as the long-legged buzzard [60], although ossification in the moa and tinamou is complete and consistent, with the trachea and syrinx of the moa also ossifying in all taxa.
Although rare among palaeognaths, ossification of the larynx, tracheal, and syrinx is common among Aves [61]. Ossification of the syrinx has been related to conferring rigidity in the syrinx as an adaptation for vocalisation, although the benefit of laryngeal and tracheal ossification remains unresolved. In comparison to mammalian counterparts, the avian trachea is well adapted to reducing chances of collapse with the presence of complete rings, and ossification is potentially a supplementary adaptation to assist with this [61]. If ossification is not an ancestral avian character linking tinamou and moa and other outgroup taxa, but a derived character, further investigation into the morphology of moa and tinamou may find drivers in call-type or behaviour related to the ossified state.
In addition, a cranial character, character 43 (state 1, articulation of the maxillary process of the nasal with the maxilla) was identified as an apomorphic character for the tinamou and moa through a review of the literature [62]. In the molecular topology, state 1 is placed unambiguously as a synapomorphy of the tinamou/moa clade; however, the character is not free of homoplasy as state 1 also occurs in Lithornis and the outgroups. In the combined data topologies, state 1 is again shared by the tinamou and moa, but not identified as synapomorphic due to the basal position of this clade within palaeognaths. Similarly to character 29 and 37 discussed above, because state 1 occurs in the sampled outgroup taxa (and also Lithornis) it is instead inferred to as the plesiomorphic state for palaeognaths, retained in tinamou and moa.
The identified link between discussed characters in the tinamou/moa clade and many outgroup taxa drove the syrinx, hyoid, and larynx data to support a tinamou/moa divergence more basal than that proposed in the molecular data topology. Although this is likely a result of characters retained from the last common ancestor between palaeognaths and neognaths, and not an indication of true descent, a recent study testing phylogenomic supertree methodologies on the avian topology, placed the tinamou/moa clade low in the palaeognath lineage [63] as the sister group to palaeognaths other than Struthio. The supertree was composed using three palaeognath phylogenetic analyses and three topological backbones, all of which prioritised molecular and genomic data and thus, morphological convergence should have had limited influence. Although no confirmation can be made, these results may indicate the divergence of tinamou and moa to be as of yet, unresolved relative to rheids in the tree.
In the molecular topology the basal positioning of the rhea and ostrich is supported by the unambiguous, unique and unreversed SHL character 35 (state 3, diamond shaped procricoid, dorsal view) and homoplasious, ambiguously optimised characters 10 (state 1, absence of the tympanum) and 39 (state 1, flattened arytenoid shape) identified as apomorphic for all other palaeognaths. Conversely, in the morphological topology, characters 35 (state 3) and 39 (state 1) optimise as unambiguous, unique and unreversed character states for the rhea and ostrich clade.