Anatomy of the forelimb musculature and ligaments of Psittacus erithacus (Aves: Psittaciformes)

Abstract

Parrots (order Psittaciformes) are a rather homogeneous group of birds that can be easily distinguished by the notably modified morphology of the skull and hindlimb. Detailed description of the forelimb morphology in these birds has never been provided, though parrots are often used as model objects in flight studies. Parrots are also considered the closest living relatives of the perching birds (Passeriformes), and thus knowledge of the wing morphology in Psittaciformes is important for understanding the evolution of the locomotor apparatus on the way to the most speciose group of birds. Here we provide a comprehensive illustrated description of the wing morphology (musculature and ligaments) of the African grey parrot (Psittacus erithacus) and compare it with several closely related taxa of the high clade Eufalconimorphae and more distantly related outgroups (based on personal dissections and literature data). We note a general similarity of the wing musculature between P. erithacus and Falconidae. A number of features common with the outgroup Columbidae supports a generally plesiomorphic structure of the forelimb in parrots as compared with the Passeriformes. Nevertheless, the wing of the Psittaciformes displays a series of structural (likely autapomorphic) modifications, which can be explained in terms of adaptations for flight with vertical body. An analysis of the anatomical data for parrots (ratio of wing elevators and highly unusual development of the M. supracoracoideus), which is based on the current experiment‐based knowledge of the flapping flight in birds, allows us to hypothesize that parrots are able to produce useful aerodynamic force during the upstroke, which is also known for pigeons and hummingbirds. This supposed ability of vertical flight and the zygodactyl foot together link the origin of parrots with the dense (likely tropical) forests.

Introduction

Parrots (order Psittaciformes) are a rather homogeneous group of birds that comprises nearly 400 species that can be easily distinguished by their notably modified morphology (Sibley & Ahlquist, 1990; Del Hoyo et al. 1997; Forshaw, 2010). The skull of parrots has an unusual hinge‐like cranio‐facial articulation (Tokita, 2003), their foot is fully zygodactyl (i.e. the outer toe is permanently reversed) as in cuckoos (Cuculidae), woodpeckers and allies (Piciformes), but the morphological basis of the zygodactyly in parrots differs from that of other birds with reversed outer toe, and thus Psittaciformes apparently evolved such toe arrangement independently (Steinbacher, 1935; Botelho et al. 2014). Surprisingly, the musculoskeletal system of parrot pectoral girdle and wing received little attention in the literature. Almost all of what we know about the forelimb anatomy of parrots results from works of the anatomists of the XIX century. Fürbringer (1888), throughout his contribution to avian morphology, studied and illustrated only the shoulder muscles of three species of parrots; and Gadow & Selenka (1891) only superficially described several muscles of the forelimb. Beddard & Parsons (1893) and Beddard (1898) studied anatomy (including myology) of several dozen genera of parrots, including Psittacus. They made some comments on the structure of several muscles and their differences among species (Beddard & Parsons, 1893; Beddard, 1898), but did not provide detailed descriptions. Evans (1996) briefly described several (but not all) wing muscles of Melopsittacus undulatus. The most recent and complete research on parrot anatomy has been devoted to the most atypical representative of the order and the only flightless parrot species, the kakapo (Strigops habroptila; Livezey, 1992). The latter publication does not provide a detailed description of the musculoskeletal system of kakapo, but still it gives detailed illustrations of the wing muscles and skeleton of the pectoral girdle as well as valuable data on the qualitative and quantitative anatomical features of the kakapo in comparison with the kea (Nestor notabilis; Livezey, 1992).

Although complete morphology of the parrot wing has thus never been comprehensively described and illustrated in detail, it represents a potentially interesting study object that may yield fruitful data for understanding morphological evolution in the avian lineage leading to the most speciose avian order Passeriformes. For a long time, ornithologists thought that parrots had no close relatives in the recent fauna (reviewed in Sibley & Ahlquist, 1990), but recent works (Suh et al. 2011; Jarvis et al. 2014; Burleigh et al. 2015) robustly place them in a clade Eufalconimorphae, which further includes perching birds (Passerifomes) and falcons (Falconiformes). Such a phylogenetic position of parrots does not contradict the fossil record (Mayr, 2014), and it is notable that early parrot‐like birds do resemble falcons in the structure of their forelimb bones (e.g. humerus; Mayr, 2011). Understanding the structure and function of the forelimb in extant parrots is thus important to reconstruct the forelimb function in the early evolution of the clade Eufalconimorphae. A good trainability of parrots makes them model objects in studies of flight kinematics (Hedrick et al. 2003, 2007; Hedrick & Biewener, 2007). The authors of the cited works admit that the wing anatomy is similar in parrots and pigeons despite the fact that the species are phylogenetically distant (Hedrick & Biewener, 2007). However, it is important to reveal which features of the parrot anatomy are unique, and which are common with other flying birds, in particular more thoroughly investigated pigeons.

We have studied forelimb musculature and arthrology of the African grey parrot (Psittacus erithacus). For the first time in the anatomical literature, we provide detailed descriptions and illustrations of each muscle and tendon of the pectoral girdle and the wing of a parrot. Our results demonstrate how unusual the wing morphology of parrots is. We found that parrots show a significant similarity with falcons in the anatomy of the wing, but differ considerably from the passerines. In addition, parrots demonstrate a unique ratio of wing elevators that we relate with their possible ability to generate aerodynamic force during the upstroke as in the hummingbirds and pigeons (Ros et al. 2011; Warrick et al. 2012).

Materials and methods

Three specimens of the African grey parrot (P. erithacus) were obtained from Severtsov Institute of Ecology and Evolution of the Russian Academy of Sciences in Moscow. For comparisons and discussion, we further dissected one specimen of the following bird species: Garrulus glandarius and Pica pica (Passeriformes), Psittacus timneh, Psittacula krameri, Poicephalus senegalensis, Cacatua ducorpsii (Psittaciformes), Falco tinnunculus and Falco vespertinus (Falconiformes), Dryocopus martius and Dendrocopos major (Piciformes), Columba livia (Columbiformes) and Coracias garrulus (Coraciiformes). We also used skeletons from the comparative osteological collection of the Borissiak Paleontological Institute of the Russian Academy of Sciences in Moscow (PIN). Fresh specimens were dissected under a Carl Zeiss Stemi SV 11 stereomicroscope. Photos were taken with Canon EOS 600D, illustrations were prepared in the Inkscape software.

After the dissection and gross anatomical study, we systematically removed muscles and carried out weight analysis. Muscle mass measurements were obtained to the nearest 0.00001 g. The data were processed in Microsoft^® Office Excel 2003 SP3. Obtained data were used for qualitative functional discussions as they allow us to confirm clearly the revealed morpho‐functional features of the parrot's wing (Table S1).

The description of ligaments is followed by short notes on their function, as there are critically few data about the anatomy and function of bird arthrology in the literature. The anatomical nomenclature follows Nomina Anatomica Avium (Baumel et al. 1993) unless otherwise stated.

Myology

Muscles of the shoulder girdle

M. rhomboideus superficialis (RS; Figs 1, 2, 3) is located dorsally on the trunk, and in its middle part is covered from above by the M. latissimus dorsi. This muscle originates from the processes spinosus of the last cervical and 1–6 thoracic vertebrae as well as the cranial edge of the synsacrum and interspinous ligaments. There is a gap in fibers at the level of the collum scapulae. The muscle inserts on the entire dorsal edge of the scapula, except the most caudal 1/9 of its length.

Figure 1.

Bones of the pectoral girdle and the wing of the African grey parrot (Psittacus erithacus), showing muscle attachments: (a) lateral view of sternum; (b) dorsal view of left coracoid; (c) ventral view of left coracoid; (d) lateral view of right scapula; (e) medial view of right scapula; (f) cranial view of right humerus; (g) ventral view of right humerus; (h) dorsal view of right carpometacarpus; (i) ventral view of right carpometacarpus; (j) dorsal view of right radius; (k) ventral view of right radius; (l) dorsal view of right humerus; (m) caudal view of right ulna; (n) cranial view of right ulna; (o) caudal view of right humerus.

Figure 2.

Dorsal view of the superficial muscles and ligaments of the left wing of the African grey parrot. Above: photograph. Below: interpretative drawing.

Figure 3.

Dorsal view of the middle muscular layer of the right shoulder girdle and shoulder of the African grey parrot. Left: photograph. Right: interpretative drawing.

M. rhomboideus profundus (RP; Figs 1 and 2) is located dorsally on the trunk and is almost completely covered by the M. rhomboideus superficialis. The M. rhomboideus profundus arises from the processes spinosus and interspinous ligaments of the last cervical and 1–6 thoracic vertebrae by an aponeurosis, and extends in the caudolateral direction ending fleshy by a narrow strip on the medial surface of the caudal half of the scapula directly under the M. rhomboideus superficialis.

M. serratus superficialis (Figs 1, 4 and 5) is located on the dorsolateral surface of the thorax and is composed of three distinct parts: cranial (pars cranialis), caudal (pars caudalis) and metapatagial (pars metapatagialis).

Figure 4.

Ventral view of the deep muscles of the right shoulder girdle of the African grey parrot. Left: photograph. Right: interpretative drawing.

Figure 5.

Lateral view of the deep muscles of the right shoulder girdle of the African grey parrot. Left: interpretative drawing. Right: photograph.

The pars cranialis (SSCr; Figs 1, 4 and 5) arises by three distinct heads tendinously and fleshy from the lateral surface of the caudal part of the last two cervicodorsal ribs and from the cranial part of the lateral surface of the dorsal segment of the first true rib. It ends by a long and wide aponeurosis on the cranial third of the ventral edge of the scapula between the medial and lateral heads of the M. subscapularis. The middle head, which arises from the last cervicodorsal rib, is significantly much more developed than the other two, and so it accounts for the bulk of the muscle.

The pars caudalis (SSca; Figs 1, 4 and 5) originates by three heads tendinously and fleshy from the lateral surface of the dorsal segments of the 1–3 true ribs and their uncinate processes. Pars caudalis forms a wide rectangular belly and ends by an aponeurosis on the ventral edge of the caudal third of the scapula. At the distal apex of the scapula, the muscle also inserts on the scapula by fibers that lay medial to the aponeurosis of insertion.

The pars metapatagialis (SSM; Fig. 5) arises slightly ventrocaudal than the pars caudalis. It originates aponeurotically by two heads from the dorsal segments of the third and fourth true ribs. Parallel muscular fibers form a long ribbon‐like belly, which ends on the metapatagium.

M. serratus pars profundus (SP; Figs 1 and 6) is a small muscle, it lies deep to the rhomboideus muscle complex. This muscle arises from the caudal edge of the dorsal part of the two cervicodorsal ribs and inserts on the medial surface of the cranial part of the caudal half of the scapula directly under the rhomboideus muscles.

Figure 6.

Dorsal view of the deep muscles and ligaments of the right shoulder girdle of the African grey parrot. Left: photograph. Right: interpretative drawing.

M. sternocoracoideus (SC; Figs 1, 4, 7 and 19) is located on the ventrolateral side of the thorax, in the area of the sternocoracoid articulation. This muscle originates on the cranial edge of the ventral half of the ventral segment of the first true rib, on the ventral surface of the medial part of the processus craniolateralis of the sternum, and on the cranial edge of the labrum externum of the sternum directly medial to the most cranial incisura costalis. It originates fleshy and ventrally also by three aponeuroses: at the most distal, medial and proximal parts of the muscle. The parallel‐fibered belly lies mostly on the dorsal side of the coracoid and inserts by both fibers and an aponeurosis on the dorsal surface of the distal part of the coracoid, as well as on the margo caudolateralis of the coracoid. The attachment site forms a distinct imprint on the coracoid (impressio M. sternocoracoidei).

Figure 7.

Ventral view of the middle muscles of the right shoulder girdle and shoulder of the African grey parrot. Left: photograph. Right: interpretative drawing.

Dorsal musculature of the shoulder joint

M. latissimus dorsi (Figs 1, 2, 3) lies most superficially on the dorsal part of the trunk. It is composed of two parts: pars cranialis and pars caudalis. Pars metapatagialis is absent.

The pars cranialis (LDCr; Figs 1, 2, 3) originates fleshy on the processes spinosus of the second and third thoracic vertebrae and interspinous ligaments. Parallel muscular fibers extend from the vertebral column to the shoulder and form a ribbon‐like muscular belly. At the shoulder region, the muscle runs as a thin ribbon between the M. scapulotriceps and M. humerotriceps, and inserts fleshy on the caudal surface of the proximal third of the humerus directly dorsal to the M. humerotriceps.

The pars caudalis (LDCa; Figs 1, 2, 3) is strongly shifted caudally: it begins tendinously on the pelvic fascia and on the crista iliaca dorsalis of the pelvis. Muscle fibers form a ribbon‐like parallel‐fibered belly, which is directed more cranially than that of the pars cranialis. The belly extends directly beneath the cranial part of the muscle, where it passes into a narrow tendon that inserts directly between the M. humerotriceps and the pars cranialis at the proximal part of the humerus.

M. scapulohumeralis cranialis is absent.

M. scapulohumeralis caudalis (SHCa; Figs 1, 2, 3 and 7) is a large muscle that is located beneath the M. latissimus dorsi on the dorsolateral surface of the trunk. It originates fleshy (dorsally) and tendinously (ventrally) from the caudal half of the lateral surface of the scapula across its entire length. The ventral superficial aponeurosis of origin extends distally underlying the proximal part of the muscular belly. The thick and elongated belly of the muscle markedly tapers distally. The muscle ends on the ventrocaudal surface of the crus dorsale fossae of the humerus directly between the muscular fibers of the caput ventrale of the M. humerotriceps mainly by a superficial aponeurosis, covering fibers of the M. scapulohumeralis caudalis from the lateral and dorsolateral sides.

M. subcoracoscapularis (M. subcoracoideus + M. subscapularis; SubCS; Figs 1 and 6, 7, 8). The M. subcoracoideus and M. subscapularis are inseparable and form a single complex (M. subcoracoscapularis) composed of four heads: caput laterale (SubCSl), caput mediale (SubCSm), caput dorsale (SubCSd) and caput ventrale (SubCSv). The fibers of the caput laterale originate on the middle third of the lateral surface of the scapula; the fibers of the caput mediale originate on the cranial half of the medial surface of the scapula. The fibers of the caput dorsale originate on the medial surface of the head of the scapula, on the caudal edge of the processus procoracoideus of the coracoid and on a small area of the medial part of the dorsal surface of the coracoid shaft directly caudal to the processus procoracoideus, as well as on the ligamentum sternoacrocoracoideum directly caudal to the processus procoracoideus; this fan‐shaped head is a part of the fibers of the medial head, with which they are tightly integrated. The caput ventrale originates on the ligamentum sternoacrocoracoideum and on the membrane between this ligament and the coracoid, it lies on the dorsal surface of the distal coracoid. The head crosses the shaft of the coracoid, and goes laterally to the humerus from the ligamentum sternoacrocoracoideum. The four heads converge on a short wide aponeurosis of insertion, which ends on the caudal part of the incisura capitis of the humerus and on the cranial surface of the tuberculum ventrale of the humerus, ventrocranial to the insertion of the M. coracobrachialis caudalis.

Figure 8.

Caudolateral view of the deep muscles of the right shoulder girdle of the African grey parrot. Above: photograph. Below: interpretative drawing.

M. deltoideus (Figs 1, 2, 3, 9, 10 and 26) is located superficially on the dorsal side of the shoulder. The muscle is composed of four separate parts: pars major, pars propatagialis cranialis, pars propatagialis caudalis and pars minor.

Figure 9.

Ventral view of the superficial muscles of the left shoulder girdle and shoulder of the African grey parrot. Left: photograph. Right: interpretative drawing.

Figure 10.

Proximal views of the humerus, showing aponeuroses of insertion of the M. pectoralis in various bird taxa: (a, d) Pica pica; (b, e) Psittacus erithacus; (c, f) Falco naumanni. Upper row: aponeuroses without muscle fibers; bottom row: the same with muscle fibers attachments shown.

The M. deltoideus pars major (DMa; Figs 1, 2, 3) is formed by only a small bundle of parallel fibers, which begins on the lateral surface of the cranial part of the scapula, directly dorsal to the processus glenoidales scapulae. It ends on the humerus dorsal to the M. latissimus dorsi.

The pars minor (DMi; Figs 1 and 3) is rather large as compared with the small pars major. It originates widely over the entire dorsocranial surface of the caput acrocoracoidei (head of the processus acrocoracoideus of the coracoid), and also on the dorsal surface of the acromion of the scapula. The parallel‐fibered muscular belly inserts on the dorsocranial surface of the proximal part of the crista deltopectoralis of the humerus.

The pars propatagialis cranialis (DPCr; Figs 1, 2, 9, 10 and 26) is the largest part of the M. deltoideus in P. erithacus. It originates fleshy on the medial part of the craniodorsal surface of the processus acrocoracoideus of the coracoid, on a small area of the dorsal edge of the cranial part of the scapula and on the medial surface of the processus acromialis claviculae. The caudal part of the muscle is connected to the middle of the dorsal scapular edge by a thin tendon (‘scapular anchor’). The wide and flattened belly extends into the propatagium: the cranial fibers insert on the ligamentum propatagialis, the caudal fibers join a wide aponeurotic sheet, which together with the tendon of insertion of the M. deltoideus pars propatagialis caudalis extends to the forearm, where it inserts on the lateral surface of the ulna. A part of the aponeurotic fibers diverges from this sheet and ends in the tendon of origin of the M. extensor carpi radialis.

Pars propatagialis caudalis (DPCa; Figs 1 and 9) originates beneath the pars propatagialis cranialis on the medial part of the craniodorsal surface of the processus acrocoracoideus of the coracoid and mainly on the inner surface of the aponeurosis of origin of the M. deltoideus pars propatagialis cranialis. It forms an elongated and diagonally oriented fusiform belly, which is superficially completely covered by the pars propatagialis cranialis. The tendon of insertion of the pars propatagialis caudalis is fused with the caudal part of the M. deltoideus pars propatagialis cranialis.

Ventral musculature of the shoulder joint

M. pectoralis (Pt; Figs 1, 7, 9 and 10) occupies most of the ventral surface of the thorax. The muscle is composed of three parts: the largest pars thoracicus and two additional ones: pars propatagialis longus and pars propatagialis brevis.

The pars thoracicus has a wide area of origin. It originates on the sternum and adjacent regions: on the caudolateral and caudal edges of the corpus sterni; on the ventral part of the lateral surface of the carina sterni; on the ventral segments of the true ribs; on the caudolateral edge of the corpus claviculae; and on the membrana sternocoracoclavicularis. The pars thoracicus ends on the crista deltopectoralis of the humerus along its entire length by two aponeuroses (lateral and medial) and fleshy but, besides this typical attachment area, the muscle also ends on the crus ventrale fossae of the humerus by an additional small aponeurosis. The lateral (external) aponeurosis inserts on the outer edge of the crista deltopectoralis, and the medial aponeurosis inserts closer to the shaft, along the base of the crista deltopectaralis. Muscle fibers insert directly on the crista deltopectoralis between the lateral and medial aponeuroses. The medial aponeurosis forms a complex and branching system, whereas the lateral one is composed of only one superficial sheet.

The system of medial aponeuroses comprises several parts (Fig. 10), which are termed in our study as follows: the main aponeurosis (aponeurosis medialis, am); superficial medial aponeurosis (aponeurosis medialis superficialis, ams); intramuscular medial aponeurosis (aponeurosis medialis intermuscularis, ami); and additional free medial aponeurosis (aponeurosis medialis accessorius, ama). The intramuscular medial aponeurosis (ami) is the most durable and extended; the superficial medial aponeurosis (ams) is rather short in general, but its middle part forms a long ‘tongue’, extending up to the proximal one‐third of the dorsal surface of the muscular belly. Additional medial aponeurosis (ama) is robust and proximodistally short, and represents a continuation of the ‘tongue’ of the superficial medial aponeurosis. It arises from the interface of all medial aponeuroses and merges to the short tendon of origin of the M. biceps brachii across its entire width. This aponeurosis inserts on the base of the crus ventrale fossae of the humerus. There are no muscle fibers attached directly to this additional aponeurosis. The later superficial aponeurosis, which inserts on the edge of the crista deltopectoralis, collects a significantly smaller amount of muscle fibers in comparison with the system of medial aponeuroses. It covers only a small portion of the distal part of the craniolateral surface of the muscular belly.

The two propatagial portions of the pectoralis muscle are represented by reduced tendon sheets – tendo propatagialis brevis and tendo propatagialis longus. They arise from the superficial ventral aponeurosis of the pars thoracicus and run to the propatagium. The tendo propatagialis longus originates on the cranial part of the thoracic part of the muscle at the level of the cranial edge of the crista deltopectoralis and joins the tendon of insertion of the M. deltoideus pars propatagialis cranialis to form the ligamentim propatagialis. The tendo propatagialis brevis originates on the cranial part of the pars thoracicus cranial and distal to the tendo propatagialis longus. Distally it merges with the tendon of insertion of the M. deltoideus pars propatagialis caudalis directly distal to its muscular belly and ends on the tendon of origin of the M. extensor carpi radialis.

M. supracoracoideus (SupC; Figs 1 and 7) is located between the corpus sterni and the M. pectoralis. It arises fleshy from the lateral surface of the membrana sternocoracoclavicularis, on the ventromedial surface of the coracoid, but mainly on the base of the carina sterni and the medial part of the corpus sterni, along the full length of the sternum right up to its caudal edge. Muscular fibers of the bipennate belly converge on an intramuscular aponeurosis, which give rise to the tendon of insertion, typically passing through the canalis triosseus and inserting on the dorsocaudal surface of the humerus just distal to the tuberculum dorsale.

M. coracobrachialis cranialis (CBCr; Figs 1 and 11) occupies a dorsocranial surface of the proximal humeral end just dorsal to the M. biceps brachii. It originates by a short tendon on the lateral surface of the caput acrocoracoidei (processus acrocoracoideus of the coracoid) and ends fleshy on the cranial surface of the proximal part of the humerus directly distal to the caput humeri and ventral to the crista deltopectoralis. The attachment site on the humerus forms a characteristic imprint (impresso M. coracobrachialis). The muscular belly is fully covered by a connective tissue sheet, which represents a cranial branch of the capsula articularis.

Figure 11.

Ventral view of the skeleton and ligaments of the left shoulder girdle of the African grey parrot. Above: photograph. Below: interpretative drawing.

M. coracobrachialis caudalis (CBCa; Figs 1, 4 and 7) is located on the ventrolateral body surface directly beneath the M. pectoralis and dorsolateral to the M. supracoracoideus. It originates fleshy and by a small aponeurosis on the lateral part of the ventral surface of the extremitas sternalis of the coracoid, on the lateral surface of the distal part of the coracoid shaft and also on the processus craniolateralis of the sternum.

The muscular fibers form a bipennate belly and converge on an intramuscular aponeurosis and on a very short superficial aponeurosis, which covers a small area of cranial surface of the muscle. The aponeuroses form a tendon of insertion that inserts on the tuberculum ventrale of the humerus dorsocaudal than M. subscapularis.

Musculature of the shoulder

M. triceps brachii (Figs 1, 2, 3, 7, 9 and 12) is a large muscle composed of two fully separate parts, M. scapulotriceps and M. humerotriceps, which cover the whole caudal surface of the humerus. M. coracotriceps is absent.

Figure 12.

Ventral view of the superficial muscles of the right shoulder, forearm and wrist of the African grey parrot. Above: photograph. Below: interpretative drawing.

The M. scapulotriceps (TBs; Figs 1, 2, 3 and 7) lies superficially on the caudal side of the humerus. The muscle begins fleshy on the dorsal branch of the ligamentum scapulohumerale and by a large tendon, which expands into a superficial aponeurosis that covers the cranial two‐thirds of the ventral surface of the muscular belly. The muscular fibers end on a distal superficial aponeurosis, which covers the distal two‐thirds of the dorsal surface of the muscle. The distal superficial aponeurosis gives rise to a long, large and flattened tendon of insertion, which passes through the sulcus scapulotricipitale of the humerus, crosses the elbow joint and inserts on the base of the processus cotylaris dorsalis of the ulna. At the level of the elbow joint, the tendon contains a sesamoid and is connected with the tendon of the M. humerotriceps by a loose sheet of connective tissue.

M. humerotriceps (TBh; Figs 1, 2, 3, 7, 9 and 12) is slightly larger than the M. scapulotriceps, it occupies the caudal and ventrocaudal surfaces of the humerus. The muscle typically begins tendinously along the entire length of the ventral part of the caudal surface of the humerus. Proximally, the muscle is divided into two heads by the insertion of the M. scapulohumeralis cranialis: the caput dorsale begins on the humerus dorsal to the fossa pneumotricipitalis, and the caput ventrale begins by a thin tendon and fleshy inside the fossa pneumotricipitalis and also ventral to it. Just caudal to the pneumatic cavity, the fibers of both heads converge on an intramuscular aponeurosis that more distally forms the tendon of insertion. This tendon and a part of muscular fibers cross the elbow joint and then attach to the proximal part of the olecranon of the ulna. A small amount of fibers at the level of the elbow joint inserts on the tendon of insertion of the M. scapulotriceps.

M. biceps brachii (BB; Figs 1, 2, 7, 9, 10 and 12) is a large spindle‐shaped muscle located on the ventrocranial surface of the humerus. The proximal part of this muscle is covered by the M. pectoralis. The M. biceps brachii typically originates by two flat tendons: the longer one (caput humerale) begins on the ventral surface of the proximal part of the processus acrocoracoideus of the coracoid, and the shorter one begins on the cranial part of the incisura capitis of the humerus. The shorter tendon (caput coracoideum) is connected with the additional medial aponeurosis (ama) of the M. pectoralis (see above). Both shorter and longer tendons merge to form a long and large common aponeurosis of origin, which is further attached to the cranial branch of the capsula articularis. The muscular belly is stretched along the humerus and consists of short muscular fibers tightly packed between the aponeurosis of origin and the aponeurosis of insertion. A free part of the aponeurosis of insertion is short and bifurcates into two flat tendons in the forearm. The weaker of the two branches ends on the tuberculum bicipitale radii, and the more developed one inserts on the tuberculum bicipitale ulnae.

Dorsal musculature of the forearm

M. extensor carpi radialis (ECR; Figs 1, 2 and 12, 13, 14) is the largest and most complex muscle of the forearm, which occupies its cranial surface. The muscle has two aponeuroses of origin, which we term the aponeurosis superficialis dorsalis proximalis (asdp) and the aponeurosis superficialis ventralis proximalis (asvp), and further two aponeuroses of insertion, the aponeurosis superficialis dorsalis distalis (asd) and the aponeurosis profundus distalis (apd). The muscular fibers are very short and form a complex belly subdivided into several small parts due to a complex aponeurotic structure of the muscle.

Figure 13.

Dorsal veiw of the superficial and deep muscles of the right forearm and wrist of the African grey parrot. Above: photograph. Below: interpretative drawing.

Figure 14.

Relative position of the M. extensor carpi radialis aponeuroses in the African grey parrot: (a) cranial view, scheme; (b) ventral view; the aponeurosis superficialis ventralis proximalis (asvp) is bent and cut off; aponeurosis superficialis dorsalis proximalis (asdp) shifted cranially; origin of the fibers from the aponeuroses is shown.

Two heads can be distinguished within the muscle: the caput ventrale and the caput dorsale, which in the proximal part are largely separated from each other. The caput dorsale originates by a wide and large aponeurosis (asdp) on the processus supracondylaris dorsalis, as well as by a separate tendon on the proximolateral surface of the epicondylus dorsalis of the humerus. This tendon distally joins the proximal dorsal aponeurosis (asdp). Additionally, this aponeurosis is also proximally connected with the distal aponeurosis of the M. deltoideus pars propatagialis cranialis. Approximately in the middle of its length, the proximal dorsal aponeurosis (asdp) sends out yet another separate tendon, which passes along the forearm and ends on the dorsal surface of the distal part of the radius, caudal to the sulcus tendinosus. The caudal part of the proximal dorsal aponeurosis (asdp) ends at the level of two‐thirds of the length of the muscle belly. In the cranial part of the muscle belly this aponeurosis continues distally and in the distal third of the muscle belly lies intramuscular.

The caput ventrale begins on the processus suprocondylaris dorsalis of the humerus more ventrally than the caput dorsale fleshy and by means of the ventral superficial aponeurosis (asv). This aponeurosis (asv) covers the proximal two‐thirds of the length of the muscular belly.

The aponeurosis profundus distalis (apd) bifurcates proximally, and thus the proximal fibers of the corresponding heads insert on its sheets. The distal fibers of both heads terminate on the distal single layer of this aponeurosis. The aponeurosis dorsalis superficialis distalis (adsd) receives fibers on the cranial surface of the distal part of the muscle. Distally, in the last third of the length of the forearm, this aponeurosis, together with the aponeurosis profundus distalis (apd), forms a large tendon of insertion that extends along the dorsocranial surface of the radius, enters the sulcus tendonosus of the radius, then crosses the intercarpal joint and inserts on the apex of the processus extensorius of the carpometacarpus.

M. supinator (Su; Figs 1 and 15) is a small muscle located in the proximal half of the dorsocranial surface of the forearm. It is partially visible cranial to the M. extensor digitorum communis, and is partially covered by that muscle. The M. supinator arises by a tendon from the epicondylus dorsalis of the humerus, which is attached ventral and distal to the point of origin of the M. extensor digitorum communis (see below). The tendon of origin gives rise to an extensive dorsal superficial aponeurosis that almost entirely covers the muscular belly. Very short muscle fibers arise from the dorsal superficial aponeurosis and terminate on the dorsocranial surface of the proximal part of the radius.

Figure 15.

Ventral view of the deep muscles of the right forearm and wrist of the African grey parrot. Above: photograph. Below: interpretative drawing.

M. ectepicondylo‐ulnaris (EU; Figs 1 and 13) lies beneath the M. extensor carpi ulnaris. The muscle arises from the dorsal surface of the epicondylus dorsalis of the humerus by a wide aponeurosis common with the M. extensor carpi ulnaris. At the level of the proximal fifth of the length of the ulna, this aponeurosis splits into two parts, and its cranial part becomes the superficial aponeurosis of the M. ectepicondylo‐ulnaris. Muscular fibers originate from the common aponeurosis of origin throughout its width directly distal than the elbow joint and terminate on the dorsocranial surface of the proximal two‐thirds of the ulna.

M. entepicondylo‐ulnaris is absent.

M. extensor carpi ulnaris (ECU; Figs 1, 13 and 16) is a superficial muscle, which is visible caudolateral to the M. extensor digitorum communis (see below). The muscle originates by means of two tendons: one tendon of origin arises from the epicondylus dorsalis of the humerus proximal to the origin site of the M. ectepicondylo‐ulnaris; the second flattened tendon is shorter, it arises from the dorsocaudal surface of the proximal edge of the ulna. The tendons of origin meet each other approximately at an obtuse angle, unite and form a broad superficial ventral aponeurosis that extends almost to the very end of the muscular belly, excluding the distal one‐eighth of its length. The fibers end on the dorsal superficial aponeurosis, which covers the distal seven‐eighths of the muscle length. Directly proximal to the wrist, the distal superficial aponeurosis forms a tendon of insertion that runs along the dorsal surface of the distal part of the ulnar shaft, passes through the incisura tendinosa (together with the tendon of insertion of the M. extensor digitorum communis), then runs along the dorsal surface of the distal end of the ulna, crosses the intercarpal joint and inserts on the apex of the processus intermetacarpalis of the carpometacarpus.

Figure 16.

Dorsal view of the skeleton and ligaments of the right wrist of the African grey parrot. Above: photograph. Below: interpretative drawing.

M. extensor digitorum communis (EDC; Figs 1, 13 and 16) is a dorsal superficial muscle, arising from the dorsocranial surface of the epicondylus dorsalis of the humerus just proximal to the origin sites of the M. ectepicondylo‐ulnaris and M. extensor carpi ulnaris. A superficial aponeurosis covers the proximal two‐thirds of the dorsal surface of the muscular belly, which is stretched along the entire forearm. The muscle fibers end on a ventral superficial aponeurosis, which covers the distal two‐thirds of the muscular belly. The tendon of insertion passes inside the incisura tendinosa (together with tendon of insertion of the M. extensor carpi ulnaris), runs along the dorsal edge of the distal end of the ulna (distal to the tendon of the previous muscle), crosses the intercarpal joint and bifurcates near the distal margin of the processus extensorius. A shorter branch ends on the proximocaudal angle of the dorsal surface of the phalanx digiti alulae. The longer branch enters the sulcus tendineus1 on the dorsal surface of the carpometacarpus, runs within it up to the distal end of the carpometacarpus, and then inserts on to the dorsoproximal angle of the phalanx proximalis digiti majoris.

M. extensor longus alulae (ELA; Figs 1, 13 and 15) is a thin muscle located between the ulna and the radius. The muscle begins with two separate heads: a larger and proximal ulnar one; and a smaller and more cranially and distally located radial one. The fibers of the ulnar head originate on the cranial surface of the proximal part of the ulna, whereas the fibers of the radial head begin on the second quarter of the radius length. The fibers of both heads converge on a distal aponeurosis, which eventually give rise to the tendon of insertion. This tendon goes to the dorsocaudal side of the radius, passes along the caudal branch of the sulcus tendinosus of the radius, and inserts on the top of the processus extensorius of the carpometacarpus directly next to the insertion of the M. extensor carpi radialis.

Ventral musculature of the forearm

M. pronator superficialis (PS; Figs 1 and 12) is a superficial muscle, stretched along the cranioventral surface of the forearm. It lies just caudal to the M. extensor carpi radialis. Its large tendon of origin originates on the ventral surface of the humerus ventral to the tuberculum supracondylare ventrale. This tendon almost immediately gives rise to an extensive ventral superficial aponeurosis that covers the entire ventral surface of the muscular belly. The belly crosses the elbow joint because some muscular fibers originate on the proximal aponeurosis just near the humerus. Fibers terminate on a dorsal superficial aponeurosis that covers the distal four‐fifths of the length of the muscle belly and which along its entire length is attached to the ventrocranial surface of the radius. The distal margin of the muscle is located approximately at the level of the middle of the forearm.

M. pronator profundus (PP; Figs 1 and 12) is much larger than the M. pronator superficialis, and lies on the ventral surface of the forearm deeper and somewhat caudal to the superficial pronator. The muscle begins by means of a short tendon that inserts on the epicondylus ventralis of the humerus distal to the place of origin of the M. pronator superficialis. The short tendon of origin expands almost immediately into a superficial aponeurosis covering the proximal two‐thirds of the caudal part of the ventral surface of the muscular belly. Muscular fibers terminate on ventrocranial and dorsocranial superficial aponeuroses, which cover most of the ventral and dorsal surfaces of the muscle belly, respectively. The radial margins of the terminal aponeuroses are combined and attached to the ventrocranial surface of the proximal half of the radius. The most proximal muscular fibers of the M. extensor longus digiti majoris originate from the dorsal surface of the dorsocranial superficial aponeurosis.

M. extensor longus digiti majoris (ELDM; Figs 1, 13 and 15, 16, 17) consists of two (proximal and distal) unequal parts. The proximal one (pars proximalis) is located deep in the forearm. Its fibers begin on the ulnar surface of the middle two‐quarters of the radius and converge on an intramuscular aponeurosis, thus forming a bipennate belly. The intramuscular aponeurosis forms a tendon of insertion, which passes between the ulna and radius on the dorsal side of the forearm. The tendon passes inside the incisura tendinosa of the ulna, separated by a low ridge from an adjacent dorsocaudally located incisura, which contains tendons of the M. extensor carpi ulnaris and M. extensor digitorum communis. The tendon of insertion then passes beneath the tendon of insertion of the M. extensor digitorum communis and comes to the dorsal surface of the carpometacarpus. Further it runs along the cranial edge of the wrist and ends at the distal edge of the phalanx proximalis digiti majoris and at the proximal margin of the phalanx distalis digiti majoris. At the level of the metacarpophalangeal joint, the tendon of insertion contains a sesamoid.

Figure 17.

Ventral view of the skeleton and ligaments of the right wrist of the African grey parrot. Above: photograph. Below: interpretative drawing.

The distal part (pars distalis) is represented by a small muscular belly located in the proximal part of the wrist. It begins by a thin tendon on the dorsal surface of the capsula articularis of the carpal joint and ends fleshy on the tendon of insertion of the pars proximalis at the level of the intercarpal joint.

M. flexor carpi ulnaris (FCU; Figs 1 and 12) is the largest and most caudal among the ventral muscles of the forearm. Its caudoventral surface is connected with the flight feathers, which even produce imprints on the muscular belly after removal of the skin. The muscle belly extends almost along the entire length of the forearm and consists of very short fibers packed between extensively developed aponeuroses. The muscle begins on the epicondylus ventralis of the humerus by a large tendon that passes into the loop of the trochlea humeroulnare (TrHU; Figs 12 and 23). The tendon of origin distally gives rise to an intramuscular aponeurosis where the fibers of the belly begin. The muscular fibers terminate on a larger caudal superficial aponeurosis that covers the caudoventral, caudal and caudodorsal surfaces of the distal five‐sixths of the muscle length and further on a less extensive cranial superficial aponeurosis. The two distal aponeuroses do not merge, but form independent distal tendons that terminate on the medial surface of the os carpi ulnare. The cranial aponeurosis produces a thinner tendon, and the caudal aponeurosis forms a larger and slightly more ventrally attaching tendon.

M. flexor digitorum superficialis (FDS; Figs 1, 12 and 17) is a thin muscle located on the ventral side of the forearm ventrocranial to the M. flexor carpi ulnaris. The muscular fibers begin on the ligamentum humerocarpale and are covered by it throughout the entire length of the forearm. The fibers terminate on a superficial dorsal aponeurosis, which covers almost completely the dorsal surface of the muscular belly, except for a small proximal segment. The tendon of insertion runs across the ventrocranial margin of the os carpi ulnare and then goes distally along the ventral surface of the carpometacarpus. On the cranial side of the distal part of the carpometacarpus the tendon runs beneath the tendon of insertion of the M. flexor digitorum profundus and inserts on the proximocranial surface of the phalanx proximalis digiti majoris.

M. flexor digitorum profundus (FDP; Figs 1, 12 and 15) is a large muscle located on the ventral surface of the forearm. Its proximal part is hidden beneath the more superficially laying muscles, and the distal part is visible cranial than the M. flexor carpi ulnaris and M. flexor digitorum superficialis. The muscle begins by one head from the proximal part of the ventrocaudal and ventral surfaces of the ulna dorsal to the attachment site of the M. brachialis and M. biceps brachii, and further from the proximal aponeurosis of the M. ectepicondylo‐ulnaris. The muscular fibers terminate on a ventral superficial aponeurosis, which covers most of the distal surface of the belly. Distally, in the last quarter of the forearm, this aponeurosis forms a large tendon of insertion that throws over the processus pisiformis of the carpometacarpus, passes along the ventrocranial surface of carpometacarpus and attaches to the proximal part of the ventrocranial margin of the phalanx distalis digiti majoris.

M. ulnometacarpalis ventralis (UV; Figs 1, 12 and 15) is a small muscle, its distal part is visible in the distal part of the forearm cranial to the M. flexor digitorum profundus. The muscle originates fleshy on the distal two‐thirds of the ventral and ventrocaudal surfaces of the ulna. The fibers end on a ventral superficial aponeurosis. A large long tendon of insertion crosses the intercarpal joint diagonally in the dorsocranial direction and inserts on the dorsocranial angle of the trochlea carpalis.

M. brachialis (Br; Figs 1, 12 and 15) is a small parallel‐fibered muscle, located deep on the ventral surface of the forearm. It originates fleshy in the ventral half of the cranial surface of the distal end of the humerus directly adjacent to the epicondylus medialis and ends fleshy on the ventrocranial surface of the proximal ulna. The sites of muscle attachment on the humerus and ulna are marked by well‐defined concavities (fossa M. brachialis of the humerus; impressio M. brachialis of the ulna).

Musculature of the wrist

M. ulnometacarpalis dorsalis (UD; Figs 1 and 12) is a small and superficially located muscle, which passes along the dorso‐caudal side of the wrist joints and for the most part lies on the dorso‐caudal side of the proximal part of the wrist. The muscle originates by a wide aponeurosis on the dorsocranial surface of the ulnar distal end. The muscular fibers mainly insert on the proximocaudal part of the os metacarpale minus, but an additional portion of fibers goes to the primaries. The distal part of the muscle adjoins tightly to the dorsal side of the proximal part of the M. flexor digiti minoris.

M. extensor brevis alulae (EBA; Figs 1, 16 and 17) is a very small muscle lying on the dorsoproximal surface of the carpometacarpus. The muscular fibers originate in a muscular depression on the dorsal surface of the base of the processus extensorius. The muscle terminates by a thin tendon on the cranial surface of the proximal part of the phalanx digiti alulae.

M. abductor alulae (AbA; Figs 12, 15 and 17) is the largest muscle of the alulae located on the ventral surface of the proximal part of the wrist. The muscle fibers originate directly from the ventral surface of the trochlea carpalis of the carpometacarpus and from the tendon of insertion of the M. extensor carpi radialis. The fibers end on a superficial aponeurosis covering the cranio‐ventral edge of the distal part of the belly. The aponeurosis forms a tendon of insertion that terminates on the ventral side of the phalanx digiti alulae.

M. adductor alulae (AdA; Figs 1, 2, 12, 11 and 15, 16, 17) is a small muscle located between the phalanx digiti alulae and the carpometacarpus. The muscular fibers originate on the cranial surface of the carpometacarpus directly near the processus pisiformes. The parallel‐fibered belly is attached to the caudal surface along the entire length of the distal quarter of the phalanx digiti alulae by a thin tendon of insertion.

M. abductor digiti majoris (AbDM; Figs 1, 13 and 17) is a relatively large muscle laying deep on the ventrocranial surface of the carpometacarpus. The muscular fibers originate on the processus pisiformes of the carpometacarpus and on the cranioventral surface of the os metacarpale majus along its entire length. The distal two‐thirds of the cranioventral surface of the muscular belly are covered by a superficial aponeurosis, which attaches to the ventrocranial edge of the proximal part of the phalanx proximalis digiti majoris.

M. interosseus dorsalis (ID; Figs 1 and 16) is a small muscle located on the dorsal side of the hand between the metacarpals (ossa metacarpale majus et minus). The muscle fibers originate on the dorsocaudal surface of the proximal two‐third of the os metacarpale majus, dorsal surface of the distal part of the processus intermetacarpalis and the dorsocranial surface of the proximal two‐thirds of the os metacarpale minus. Muscle fibers converge on an intramuscular aponeurosis that passes through the entire muscle belly, forming a bipennate belly. The tendon of insertion passes within the sulcus interosseus of the carpometacarpus, extends to the dorsal side of the proximal part of the phalanx distalis digiti majoris, crosses the phalanx distally in the cranial direction and inserts on the cranial surface of the proximal part of the phalanx distalis digiti majoris.

M. interosseus ventralis (IV; Figs 1, 12, 15, 17 and 18) is located on the ventral side of the wrist and completely fills the intercarpal space (spatium intermetacarpale). The muscle fibers originate on the ventrocaudal side of the os metacarpale majus throughout its extent, on the proximal two‐thirds of the ventrocranial side of the os metacarpale minus, as well as on the adjacent ventral surface of the carpometacarpus. The muscle fibers converge on a distal intramuscular aponeurosis forming a bipennate belly. The tendon of insertion passes to the dorsal side of the hand, bypassing the M. interosseus dorsalis from the caudal side and runs along the caudal edge of the dorsal surface of the phalanx proximalis digiti majoris. It ends on the caudal side of the distal margin of the phalanx proximalis digiti majoris.

Figure 18.

Ventral view of the skeleton and deep layer of the aponeurosis ventralis carpalis of the right wrist of the African grey parrot. Above: photograph. Below: interpretative drawing.

M. flexor digiti minoris (FDM; Figs 1 and 16) is located on the caudal surface of the carpometacarpus. The muscle fibers originate on the caudal side of the carpometacarpus thoughout its entire length. The belly narrows distally, and its fibers terminate on a distal tendon, which is attached to the caudomedial angle of the phalanx digiti minoris.

M. flexor alulae (FA; Figs 1, 17 and 18) is located at the proximal part of the wrist. The muscle originates on the proximal part of the ventrocranial surface of the carpometacarpus fleshy and by three aponeuroses. The most cranial proximal aponeurosis is wide and originates on the proximal part of the trochlea carpalis directly distal to the facies articularis ulnocarpalis and on the proximal part of the ventral surface of the base of the processus extensorius of the carpometacarpus. This aponeurosis covers the proximal part of the ventrocranial surface of the muscular belly. A more caudal aponeurosis of origin begins at the ventral margin of the facies articularis ulnocarpalis, and the most caudal aponeurosis begins on the cranial surface of the processus pisiformis of the carpometacarpus. Muscle fibers converge on a single aponeurosis of insertion that attaches to the ventrocaudal angle of the proximal part of the phalanx digiti alulae. One bundle of fibers inserts directly on the bone.

Arthrology

Sterno‐coraco‐clavicular articulation

Membrana sternocoracoclavicularis (MStCrCl; Figs 4, 7, 11, 19 and 20) is an unpaired structure, stretched between the sternum, furcula and coracoids. It is attached to the entire margo cranialis of the carina sterni, to the spina externa of the sternum, to the caudal surface of the furcular rami and to the acrocoracoid processes of the coracoids.

Figure 19.

Lateral view of the skeleton and ligaments of the right shoulder girdle of the African grey parrot. Left: interpretative drawing. Right: photograph.

Figure 20.

Dorsolateral and caudal view of the skeleton and ligaments of the right shoulder girdle of the African grey parrot. Left: photograph. Right: interpretative drawing.

Ligamentum sternoacrocoracoideum (LStAc; Figs 11, 19 and 20) is closely connected with the membrana sternocoracoclavicularis. This ligament is distinguished from the former by a denser bundle of fibers that are positioned along the dorsal edge of the membrane. This large long ligament begins at the top of the spina externa of the sternum and on the dorsal surface of the angulus medialis of the coracoid. The ligament ends on the processus acrocoracoideus of the coracoid.

Lig. sternoprocoracoideum (LStPrc; Fig. 19) is a dorsolateral portion of the ligamentum sternoacrocoracoideum. It ends on the processus procoracoideus of the coracoideum.

Lig. acrocoraco‐procoracoideum (LAcPrc; Figs 11 and 19) is an extension of the ligamentum sternoacrocoracoideum that is stretched between the processus procoracoideus and the processus acrocoracoideus of the coracoid. The presence of this ligament is a characteristic feature of birds with a pronounced processus procoracoideus (Baumel et al. 1993).

Lig. intercoracoideum (LIC; Fig. 20) represents a caudolateral extension of the lig. sternoacrocoracoideum. It connects apexes of the anguli mediales of the two coracoids. The dorsal most fibers of this ligament merge with the proximal part of the lig. sternoacrocoracoideum.

Lig. sternocoracoideum internum (LStCI; Fig. 20) binds together the sternum and the coracoids at their junction point. This paired ligament begins lateral to the rostrum sterni on the labrum internum of the sternum and terminates on the dorsal surface of the shaft of the coracoid directly above the junction of the sternum and coracoid along the entire length of their articulation. A short lig. sternocoracoideum externum (LStCE; Figs 4 and 19) also binds the two coracoidea at the cranial point of the labrum externum of the sternum.

Coraco‐scapulo‐humeral articulation

Lig. coracoscapulare interosseum (LCSI; Fig. 21) joins facies articularis humeralis of the scapula and coracoids, and forms the internal wall of the glenoid cavity (cavitas glenoidalis).

Figure 21.

Lateral view of the right coraco‐scapular articulation of the African grey parrot. Left: interpretative drawing. Right: photograph.

Lig. coracoscapuloclaviculare*2 (LCrSCl; Figs 20 and 22) is a very short continuous ligament connecting acromion of the scapula, processus acromialis of the clavicula, and the processus acrocoracoideus of the coracoid. The longest and lateral most part of this ligament passes between the coracoid and the scapula.

Figure 22.

Dorsal view of the skeleton and ligaments of the left shoulder girdle of the African grey parrot. Above: photograph. Below: interpretative drawing.

Lig. acrocoracohumerale (LAcH; Fig. 11) is the largest ligament in the humeral region, it is detached from in the capsula articularis. It goes from the dorsolateral surface of the processus acrocoracoideus to the ventral part of the intumescentia humeri of the humerus, forming a deep sulcus at the attachment point (sulcus transversus).

Lig. scapulohumerale (LScH; Figs 6, 8, 20 and 22) is the largest ligament on the dorsal side of the shoulder joint. The ligament begins on the caudolateral side of the processus glenoidalis of the scapula, and then diverges dorsoventrally producing dorsal (LScHd) and ventral (LScHv) branches. The dorsal branch terminates on the caudal surface of the humerus directly distal to the caput humeri on the dorsal edge of the fossa pneumotricipitalis dorsalis. It is tightly fused with the articular capsule in its distal part (actually representing a thickening of its dorsolateral part). The attachment site of the ligament on the humerus is marked with a small distinct crest. The ligament limits pronation of the humerus, most effectively in its protracted position. The ventral branch terminates on a bony bridge separating the incisura capitis and the sulcus transversus. It limits ventral displacement of the caput humeri when the wing is abducted as well as supination of the humerus when the wing is lifted.

In addition to the last two above‐described ligaments, there is an extensive tightly arranged tissue within the capsula articularis (CAO; Figs 6, 8, 11, 20 and 22), which begins on the acrocoracoid process of the coracoid and covers the caput humeri of the humerus. This tightly arranged tissue is divided into two branches: the smaller one is located dorsally and extends to the dorsal surface of the base of the crus dorsale fossae of the humerus, it fuses with the lig. scapulohumerale in the place of their intersection. The second branch extends along the cranial surface of the proximal humerus, broadly fastening along the transverse groove from the attachment point of the lig. acrocoracohumeralis to the crista deltopectoralis. The cranial branch forms a closed pocket in which the M. coracobrachialis cranialis is located.

The cranial and caudal walls of the capsula articularis do not have pronounced compactions and thus are the weakest parts of the capsula. The dorsal and ventral walls, on the other hand, have prominent compactions that limit cranio‐caudal displacement of the caput humeri and thus set the main axis of rotation in the shoulder joint.

Elbow joint

Lig. collaterale dorsale (LCDCub; Figs 23 and 24) originates on the dorsal surface of the epicondylus dorsalis of the humerus directly proximal to the M. extensor carpi ulnaris and ends on the dorsal surface of the proximal part of the radius. The distal attachment site is marked with an unclear impression. In its distal part, the ligament is ventrally fused with the capsula articularis cubiti.

Figure 23.

Ventrocranial view of the skeleton and ligaments of the elbow joint of the right wing of the African grey parrot. Left: photograph. Right: interpretative drawing.

Figure 24.

Dorsocaudal view of the skeleton and ligaments of the elbow joint of the right wing of the African grey parrot. Above: photograph. Below: interpretative drawing.

Lig. craniale cubiti (LCrCub; Fig. 23) represents a slightly differentiated cranial part of the capsula articularis cubiti. It begins on the humerus between the condyles directly distal to the fossa M. brachialis and terminates mainly on the radius, but one bundle of fibers also ends on the ulna. The ligament blocks supination of the radius when the elbow joint is fully flexed.

Lig. radioulnare transversum (LRUT; Figs 15, 23 and 24) is a broad, dense and flat ligament connecting the dorsocaudal surface of the proximal part of the radius and the cranial surface of the proximal part of the ulna. The fibers are oriented strictly perpendicular to the longitudinal axes of the forearm bones. On the both bones there are clear imprints of the ligament insertion. This ligament controls the proximodistal slip of the radius relative to the ulna. It is also tightened when the radius is pronated.

Lig. collaterale ventrale (LCVCub; Figs 15 and 23) is the largest ligament of the elbow joint. It begins strictly on the cranial surface of the tuberculum supracondylare ventrale and extends to the ventrocranial surface of the proximal ulna, where it is attached to the tuberculum lig. collateralis ventralis.

Meniscus radioulnaris (MenRU; Fig. 24) is a dense intraarticular ligament, originating mainly on the dorsal surface of the proximal part of the radius and ending on the proximodorsal edge of the crista intercotylaris of the ulna. The base of the meniscus is connected with the dorsal part of the capsula articularis cubiti. The meniscus covers an area between the condylus dorsalis of the humerus and the cotyla dorsalis of the ulna, and its edge also contributes to the dorsal part of the elbow joint.

Joints of the wrist

Lig. radioulnare interosseum (LRUI; Fig. 17) is a short ligament connecting the distal ends of the ulna and radius. It begins from the depressio radialis of the ulna and ends at the depressio ligamentosa of the radius.

Lig. ulno‐ulnocarpale proximale (LUUcP; Fig. 17) is located more proximal then the previous one, and is oriented almost strictly transversely with respect to the ulna. The ligament begins on the ventral surface of the distal ulna at the base of the tuberculum carpale in a proximity to the condylus ventralis ulnae and ends on the proximal surface of the dorsocaudal angle of the os carpi ulnare. The attachment area on the ulna is marked with a clear imprint.

Lig. ulno‐ulnocarpale distale (LUUcDi; Fig. 17) is a very large ligament that begins on the ventral surface of the tuberculum carpale of the ulna and ends on the cranial surface of the crus longum of the os carpi ulnare. The ligament fixes the os carpi ulnare when the wrist is extended.

Lig. ulno‐radiocarpale interosseum (LURcI; Fig. 17) is a distal extension of the lig. radioulnare interosseum. The ligament lies deeply within the carpal articulation, it begins from the depressio radialis of the ulna and ends on the os carpi radiale. It restricts distal mobility of the os carpi radiale (Baumel et al. 1993).

Lig. ulno‐metacarpale ventrale (LUMcV; Fig. 17) is a long ligament that originates on the cranial surface of the distal end of the ulna directly caudal to the origin site of the lig. radioulnare interosseum. This ligament runs along the os carpi radiale, caudal to it and terminates on the ventral surface of the proximal epiphyses of the carpometacarpus proximal to the processus pisiformis. The ligament is integrated with the lig. ulno‐radiocarpale interosseum in its proximal part, distally the former ligament is fused with the caudal part of the lig. radiocarpo‐metacarpale ventrale.

Lig. radio‐radiocarpale craniale (LRRcCr; Fig. 17) is a short ligament located on the cranial edge of the junction between the radius and the os carpi radiale. It begins on the cranial edge of the tuberculum aponeurosis ventralis of the radius and ends on the proximal edge of the cranial surface of the os carpi radiale.

Lig. radio‐radiocarpale dorsale (LRRcD; Fig. 16) begins by two branches from the cranial and caudal sides of the sulcus tendinosus of the radius. Caudally the two branches fuse to form a connective‐tissue sheet, covering the dorsal part of the wrist joint. It ends on the proximal edge of the dorsal surface of the os carpi radiale.

Lig. radio‐radiocarpale caudale* (LRRcCa; Fig. 16) is a large ligament, which begins near the caudal margin of the dorsal surface of the distal end of the radius. The origin site is marked by a distinct short crest. The ligament ends on the proximocaudal angle of the os carpi radiale.

Lig. radio‐metacarpale ventrale (LRMcV; Fig. 17) is the longest ligament of the wrist. It begins on the cranial edge of the tuberculum aponeurosis ventralis of the radius directly caudal to the lig. radio‐radiocarpale craniale. The ligament crosses the os carpi radiale and ends at the proximal surface of the apex of the processus pisiformis of the carpometacarpus. A bundle of fibers from the cranial part of the ligament passes to the ventral surface of the os carpi radiale.

Lig. menisco‐metacarpale is a short ligament connecting the distal part of the ventral surface of the meniscus intercarpalis and the ventral surface of the trochlea carpalis of the carpometacarpus, attaching directly proximal than the processus pisiformis.

Lig. radiocarpo‐metacarpale craniale (LRcMcCr; Fig. 17) is a short and large ligament, which begins on the ventral surface of the distocranial angle of the os carpi radiale, and terminates on the cranial part of the ventral surface of the proximal epiphysis of the carpometacarpus directly near the cranial edge of the trochlea carpalis.

Lig. radiocarpo‐metacarpale dorsale (LRcMcD; Fig. 16) begins at the proximal caudal edge of the dorsal surface of the os carpi radiale, and terminates on the cranial part of the dorsal surface of the proximal epiphysis of the carpometacarpus directly proximal to the origin site of the M. extensor brevis alulae. The distal part of the ligament is connected with the meniscus of the carpal joint.

Lig. radiocarpo‐metacarpale ventrale (LRcMcV; Fig. 17) is a very large ligament, located directly caudal to the lig. radio‐metacarpale ventrale. The ligament begins on the ventral surface of the proximal part of the os carpi radiale and ends on the carpometacarpus directly caudal to processus pisiformis. A large convexity is present at the insertion area; the cranial surface of this convexity is fused with the base of the processus pisiformis.

Lig. ulnocarpo‐metacarpale ventrale (LUcMcV; Fig. 17) is a very large fan‐shaped ligament, which begins on the apex and the cranial surface of the crus longum of the os carpi ulnare adjacent to the apex. The ligament broadens caudally and inserts extensively to the ventrocaudal surface of the proximal carpometacarpus, from the distal part of trochlea carpalis to the base of the os metacarpale minus.

Lig. ulnocarpo‐metacarpale dorsale (LUcMcD; Fig. 16) runs from the dorsocranial surface of the crus breve of the os carpi ulnare and inserts on the dorsal surface of the proximal epiphysis of the carpometacarpus directly caudal to the attachment site of the M. extensor brevis alulae. Proximally the ligament is fused with the meniscus intercarpalis.

Lig. obliquum alulae (LOA; Fig. 17) begins on the dorsolateral surface of the processus extensorius and runs to the ventroproximal surface of the phalanx digiti alulae.

Lig. collaterale ventrale 1 (LCV1; Figs 15 and 17) consists of two separate parts. The craniolateral part runs from the ventrocranial surface of the distal end of the carpometacarpus, caudally it goes diagonally and attaches to the ventroproximal surface of the phalanx proximalis digiti majoris. The thinner and longer caudomedial part begins directly caudal to the previous one, it runs diagonally and inserts on the ventrocaudal surface of the proximal part of the phalanx proximalis digiti majoris. Both ligaments are tightened during supination and abduction of the phalanx.

Lig. collaterale craniale*1 (LCCr1; Fig. 17) is a short ligament that originates on the ventrocranial surface of the distal end of the carpometacarpus and inserts on the ventrocranial surface of the proximal end of the phalanx proximalis digiti majoris.

Lig. collaterale ventrale 2 (LCV2; Fig. 17) is a short and wide ligament, which begins on the caudal part of the ventral surface of the distal end of the carpometacarpus and inserts on the proximal part of the ventral surface of the phalanx digiti minoris.

Lig. collaterale dorsale (LCD; Fig. 16) is a narrower ligament, which is longer than its ventral counterpart. It begins on the dorsal surface of the distal part of the os metacarpale majus and terminates on the cranial portion of the dorsal surface of the phalanx digiti minoris.

Lig. collaterale ventrale 3 (LCV3; Fig. 17) is a short and wide ligament, which begins on the cranial part of the ventral surface of the distal end of the phalanx proximalis digiti majoris and inserts to the proximocaudal edge of the phalanx distalis digiti majoris. The ligament is tightened by dorsal abduction (over‐extension) of the phalanx distalis digiti majoris.

Lig. collaterale craniale 2 (LCCr2; Figs 15 and 17) is a thin ligament, which begins on the ventrocranial surface of the distal end of the phalanx proximalis digiti majoris. It extends distocaudally and inserts on the middle of the ventral surface of the phalanx distalis digiti majoris distal to the area of insertion of the lig. collaterale ventrale. The ligament is tightened by adduction of the phalanx distalis digiti majoris.

Aponeurosis ventralis carpalis (ApVC; Figs 12, 15 and 18) is a complex multi‐level aponeurotic structure. The most superficial part of it, the aponeurosis propatagialis carpalis* (ApVCpc; Fig. 18), arises from the distal part of the ligamentum propatagialis; in the place of its separation the ligamentum propatagialis has a cartilaginous sesamoid. A small superficial portion of the aponeurosis propatagialis carpalis inserts on the ventrocaudal edge of the proximal end of the phalanx digiti alulae. A larger part of fibers, represented by a thin wide layer (Fig. 15), is directed perpendicular to the fibers of the ligamentum propatagialis and joins a deeper part of the aponeurosis ventralis carpalis proximal to os carpi ulnare.

The larger and deeper fan‐shaped part of the aponeurosis ventralis carpalis begins on the ventral surface of the distal part of the radius and spreads out from the cranial edge of the carpometacarpus, sending digitationes remigales to the primary and several distal secondary feathers (ApVCdr; Fig. 12). A proximal part of the aponeurosis is fused with the distal part of the ligamentum humerocarpale and with the aponeurosis ventralis antebrachii at the level of the os carpi ulnare.

A small bundle of fibers is released from the aponeurosis propatagialis carpalis and loosely attaches to the distal edge of the ventral surface of the crus longum of the os carpi ulnare directly distal to the connection with the ligamentum humerocarpale (Fig. 18). The ligamentum radio‐metacarpale longus* (ApVCrml; Figs 12 and 15), which is represented by a denser fiber bundle, can be distinguished as a part of this aponeurosis. The ligament originates on the distal end of the radius and is divided into cranial (ApVCrmlCr) and caudal (ApVCrmlCa) branches at the level of the os carpi ulnare. The cranial branch of the ligament is directed towards the ventrocranial surface of the os metacarpale majus. Another small cranioproximally directed ligament emerges from the cranial edge of the cranial branch of the ligamentum radio‐metacarpale longus and inserts on the caudal edge of the ventroproximal surface of the phalanx digiti alulae directly next to the above‐mentioned fiber bundle of the superficial portion of the aponeurosis ventralis carpalis. The cranial branch of the ligamentum radio‐metacarpale longus holds the tendons of insertion of the flexor muscles that correspond to the retinaculum flexorium sensu Baumel et al. (1993). The caudal branch of the lig. radio‐metacarpale longus goes to the caudoventral surface of the os metacarpale minus and stretches along the caudal edge of the ventral surface of the carpometacarpus, sending the digitationes remigales to the primary and several distal secondary feathers. The ligament becomes thinner and ends on the caudal edge of the carpometacarpus. The caudal branch together with the more proximal and caudal parts of the aponeurosis ventralis carpalis correspond to the aponeurosis ulnocarporemigalis sensu Baumel et al. (1993; ApVCau; Fig. 12).

Lig. radio‐metacarpale longus continues aponeurotically from the cranial branch towards the caudal branch, covering the spatium intermetacarpale.

Ligamenta accessoria alae

Lig. humerocarpale (LHC; Figs 12 and 18) is a large and long ligament, which begins on the epicondylus ventralis of the humerus. The ligament crosses the entire forearm and merges with the aponeurosis ventralis carpalis at the level of the proximal edge of the carpal joint.

Lig. propatagialis (LPr; Figs 2, 9 and 12) is a large elastic ligament located in the cranial edge of the propatagial skin membrane. The ligament is formed proximally by the M. deltoideus pars propatagialis cranialis and M. pectoralis pars propatagialis. It crosses the forearm and inserts on the ventral half of the cranial surface of the processus extensorius of the carpometacarous. Just before the point of insertion, the ligament has a cartilaginous sesamoid, which roughly corresponds with the origin of the aponeurosis propatagialis carpalis.

Aponeurosis ventralis antebrachii and aponeurosis dorsalis antebrachii (Fig. 2) cover the ventral and dorsal muscles of the forearm, respectively. They are cranially attached to the radius and caudally they are attached to the secondary feathers. The aponeurosis dorsalis antebrachii is fused with the proximal dorsal aponeurosis (asdp) of the M. extensor carpi radialis proximally.

Discussion

Although the flight of birds is being actively studied (Hedrick et al. 2003, 2007; Videler, 2006; Hedrick & Biewener, 2007), the wing anatomy of many bird taxa remain undescribed. Up until now, a thorough anatomical description of parrot wing has not been published. There are only fragmentary data (Fürbringer, 1888; Beddard & Parsons, 1893; Beddard, 1898; Livezey, 1992; Evans, 1996) that suggest that parrots are characterized by a number of specific features in the structure of their wings. The arthrology of bird wings has been studied in a very limited number of species (Baumel et al. 1993; Sych, 1999), although ligaments may play a critical role in the wing function (Baier et al. 2007). The detailed anatomical description of the African grey parrot (P. erithacus) wing, presented above, thus substantially fills the gap in our knowledge of the anatomy of Psittaciformes, and further significantly contributes to the comparative anatomy of the articular ligaments of the bird wing.

Comparative anatomical notes

For the comparative anatomical analysis presented below, we consider only variable parts of the forelimb anatomy, and do not discuss or compare features that show broad similarity in most or all bird taxa (George & Berger, 1966). For comparisons with parrots, we focus mostly on falcons and passerines because molecular data strongly support the clade Eufalconimorphae that includes parrots, falcons and passerines (Suh et al. 2011; Jarvis et al. 2014; Burleigh et al. 2015). Forelimb anatomy of falcons was consulted after Meyers (1992, 1996) and personal observations on F. tinnunculus. Data on the diversity of the forelimb myology in passerines are taken from George & Berger (1966). We also make comparisons with pigeons as an outgroup, and in most cases we compare the anatomy of P. erithacus with that of C. livia (George & Berger, 1966; Dial, 1992a; Vazquez, 1995), except when more generalized data on pigeons are available (George & Berger, 1966). Comparison with pigeons is justified as morphological data indicate a close similarity between parrots and pigeons (likely plesiomorphic), which was previously treated as an evidence of their close relationships (Bock, 1969). Additionally, pigeons and parrots are both model objects in studies of bird flight (Dial, 1992a,b; Poore et al. 1997a; Hedrick et al. 2003, 2007; Hedrick & Biewener, 2007), and thus understanding their anatomical differences may be valuable for investigating the diversity and evolution of bird flight in general. Our results show that, among other Eufalconimorphae, the anatomy of the parrot wing (musculature) is rather similar to that of falcons and at the same time significantly differs from passerines.

We extend our data on the forelimb myology of P. erithacus to all parrots when previously published data (Fürbringer, 1888; Beddard & Parsons, 1893; Beddard, 1898; Livezey, 1992; Evans, 1996) support such notions. We have observed very limited structural variations in the forelimb musculature of various parrot species, both dissected by us (P. erithacus, P. timneh, P. krameri, P. senegalensis, C. ducorpsii) and based on literature data (including the flightless kakapo). The observed structural differences are as follows: M. scapulohumeralis cranialis is absent in P. erithacus and in C. ducorpsi, but is present (although significantly reduced) in P. krameri, P. timneh, P. senegalensis, S. habroptila; M. latissimus dorsi pars metapatagialis is present in S. habroptila, N. notabilis and M. undulatus, but is absent in all species studied by us. According to published data (Beddard, 1898; Evans, 1996), M. deltoideus major is absent in Psittacula and in M. undulatus, but we, however, have found it in P. krameri (and all other studied species) where it was comparable in size to those in P. erithacus. M. deltoideus pars major and M. deltoideus pars minor are each divided into two parts in S. habroptila and N. notabilis, but we did not find any separation of these muscles in other studied species.

There is a good number of features (here and below we consider variable features only) that demonstrate myological similarities between P. erithacus, falcons and pigeon, but distinguish all of them from passerines (George & Berger, 1966; Dial, 1992a; Meyers, 1992, 1996): the bellies of mm. rhomboideus superficialis and profundus almost completely overlap, M. rhomboideus superficialis extends along the most cranial margin of the scapula; all four heads of M. subcoracoscapularis (M. subcoracoideus + M. subscapularis) insert on the humerus via a single tendon of insertion; caput ventrale and partly caput dorsale of M. subcoracoscapularis originate not on the coracoid, but on the membrana sternocoracoclavicularis; one of the mm. deltoideus pars propatagialis in pigeons and falcons is large, although not to the extent that it is in parrots (in Falco mexicanus, the size of this muscle is externally comparable to the condition in parrots); M. scapulotriceps has only one head and originates on the glenoid process of the scapula; M. extensor longus alulae originates by radial and ulnar heads; M. extensor longus digiti majoris in P. erithacus, as in pigeons and pygmy falcons (Polihierax) but not Falco, has two unequal heads: the main proximal one in the forearm and the small distal one in the wrist; M. ulnometacarpalis dorsalis is well developed, sends muscle fibers to the primary feathers, although the muscle in P. erithacus does not have two bellies, as are present in pigeons and falcons; a complete set of six muscles controlling the fingers is presented; however, in pigeons these muscles are more hypertrophied – for example, M. abductor alulae consists of two bellies, M. flexor alulae has two large heads, and M. abductor digiti majoris is also significantly hypertrophied, while in P. erithacus and falcons the muscles are weaker and do not have additional bellies or heads.

The features that are common for P. erithacus and falcons but not C. livia are as follows: the presence of several heads3 of the M. rhomboideus superficialis (Meyers, 1992); stable presence of M. latissimus dorsi pars caudalis and its origin on the synsacrum (George & Berger, 1966; Meyers, 1992); M. flexor digitorum superficialis begins fleshy on the Lig. humerocarpale (Meyers, 1996), as well as the insertion of the M. flexor digitorum superficialis on the phalanx proximalis digiti majoris. Among the features common with falcons, which are rare in other birds, it is worth mentioning the extensive area of origin of the M. sternocoracoideus on the ribs and sternum (George & Berger, 1966; Meyers, 1992). Two of the mentioned similarities between P. erithacus and falcons are further present in passerines and thus may constitute derived features of Eufalconimorphae: the presence of the M. latissimus dorsi pars caudalis and the attachment of the M. flexor digitorum superficialis on the phalanx proximalis digiti majoris.

However, the caudal part of the former muscle is still absent in some species of studied passerines (George & Berger, 1966). Similarity with passerines is also pronounced in the presence of the M. pectoralis pars propatagialis longus and brevis (George & Berger, 1966).

There is also a number of features that are common to P. erithacus and C. livia, but are not characteristic of either falcons or passerines and thus some of them likely represent plesiomorphic conditions for parrots: M. serratus pars profundus has two heads (George & Berger, 1966); large M. supracoracoideus (Hartman, 1961; George & Berger, 1966; personal observation); M. ectepicondylo‐ulnaris originates by a tendon of origin common with the M. extensor carpi ulnaris (George & Berger, 1966). M. scapulohumeralis cranialis is absent in pigeons and P. erithacus; however, this is not a stable feature of all parrots (George & Berger, 1966; Livezey, 1992; Evans, 1996; personal observations).

The presence of two heads of the M. extensor carpi radialis can also be attributed to the features common for pigeons and parrots, as both heads are very well pronounced in contrast to the falcons and passerines in which they are less distinguished (George & Berger, 1966).

There is a number of myological features of P. erithacus that distinguish it from all other bird species described in the literature: the M. sternocoracoideus has an extensive area of origin on the ribs and sternum; the M. supracoracoideus is actually the only elevator of the wing as the M. deltoideus pars major is drastically reduced; the M. deltoideus pars propatagialis cranialis is very large, attached with the scapular anchor and completely fused distally with the M. deltoideus pars propatagialis caudalis; the M. flexor carpi ulnaris does not have a distinct subdivision into two bellies, although both distal tendons are present; complete absence of the M. expansor secundariorum and M. scapulohumeralis cranialis.

Some of the above‐mentioned features require a more detailed discussion. In most birds, the M. subcoracoscapularis, which was described by Fürbringer (1902) as a single complex muscle, is divided into two separate muscles, the M. subcoracoideus and M. subscapularis, each typically consisting of two heads (George & Berger, 1966). We describe both M. subcoracoideus and M. subscapularis in P. erithacus as a single muscle (M. subcoracoscapularis) due to the fact that the transition of the fibers from the M. subcoracoscapularis caput laterale to the M. subcoracoscapularis caput dorsale is very smooth and it is impossible to find a clear boundary between them and, in addition, the fibers of four muscle heads converge on one aponeurosis of insertion. The heads of M. subcoracoscapularis in P. erithacus correspond to the heads of individual muscles in other birds, but distally the boundary between them cannot be traced. The caput dorsale and ventrale of M. subcoracoscapularis in P. erithacus differ from the M. subcoracoideus in most birds in their considerable extent, medio‐lateral orientation of the bellies and proximal attachment to the membrana sternocoracoclavicularis but not to the coracoids, as typical in birds. This caput most likely corresponds to the ventral head of the M. subcoracoideus of most birds. Despite the fact that the fibers of the M. subcoracoideus and M. subscapularis in Falco sparverius also converge to a common aponeurosis and are attached together to the humerus, Meyers (1992) describes them as two separate muscles.

The largest muscle of birds, the M. pectoralis, consists of the main portion (pars thoracicus) and may have several small variably developed portions, pars propatagialis and pars abdominalis, which can be further divided into several parts (George & Berger, 1966; Baumel et al. 1993). Among these additional portions, in P. erithacus there are two portions of the pars propatagialis, and both are degenerated into tendons. Small portions of the pectoral muscle have a similar anatomy in passerines, but in falcons the M. pectoralis pars propatagialis longus is represented by a large muscular belly (data for F. sparverius, Meyers, 1992), and in pigeons the M. pectoralis pars propatagialis longus and M. pectoralis pars propatagialis brevis are represented by small, separated bundles of fibers separated from the M. pectoralis pars thoracicus (Dial, 1992a).

Probably the most striking feature of the wing musculature in parrots (see also Fürbringer, 1888; Beddard, 1898) is the relative development of the wing elevators (M. deltoideus, M. supracoracoideus). The composition of the M. deltoideus is variable in birds (George & Berger, 1966). Psittacus erithacus has a complete set of parts, but their ratio is extremely atypical. The M. deltoideus pars major, typically large in birds, in parrots is reduced to a tiny parallel‐fibered muscular belly and is unlikely to have any significant effect on wing elevation. According to the published data, it is also very small or absent in other parrots (Beddard, 1898; George & Berger, 1966; Evans, 1996). The M. supracoracoideus, on the contrary, is hypertrophied and its origin site occupies a very broad area on the sternum. The M. supracoracoideus occupies the sternum up to its caudal margin, and not just the cranial part, as in most birds (George & Berger, 1966). For example, in falcons, this muscle is small and is attached only to the cranial half of the sternum (Meyers, 1992). In passerines, the M. supracoracoideus also occupies only the cranial part of the sternum (Hudson & Lanzillotti, 1955; George & Berger, 1966), although its size may be quite large (Dial et al. 1991). The ratio of wing elevators, which is characteristic of parrots, does not occur in passerines, falcons or pigeons but, interestingly, it is typical for birds with very specialized forelimb function, such as hummingbirds (George & Berger, 1966; Warrick et al. 2012) and penguins (Schreiweis, 1982). Another part of the deltoid complex (pars propatagialis cranialis) is the largest part of the deltoid complex in P. erithacus as a result of its significant hypertrophy and not only because of the reduction of the M. deltoideus pars major. George & Berger (1966) note that the pars propatagialis (‘mm. tensor patagii longus et brevis’) is hypertrophied in pigeons and parrots, and Beddard (1898) notes that this portion in parrots is the largest among all other birds. A similar hypertrophy of one of the mm. deltoideus pars propatagialis (‘mm. tensor patagii longus et brevis’) is apparently present in F.o mexicanus (see fig. IX. 28 in George & Berger, 1966). However, the hypertrophied M. deltoideus pars propatagialis cranialis is not accompanied by a reduction of the M. deltoideus major neither in pigeons, nor in falcons, which is large and complex in these two groups (Dial, 1992a; fig. 3, George & Berger, 1966).

Although some birds (e.g. pigeons and some falcons) have the structure of the mm. propatagialis of the deltoid complex, which is generally similar to that of parrots (George & Berger, 1966; Meyers, 1992; personal observations), a significant difference in the topography of these muscles exists between parrots and other birds. Previous authors described a common muscle belly of the mm. deltoideus propatagialis in parrots (Fürbringer, 1888; George & Berger, 1966), as well as in pigeons and some falcons (George & Berger, 1966). However, we have found two separate muscle bellies in all studied parrots, pigeons and falcons, though these bellies are variably integrated to each other (especially well integrated in C. ducorpsii). In parrots, the M. deltoideus propatagialis is represented by two parts: the pars propatagialis cranialis (which inserts to the lig. propatagialis and to the forearm) and the pars propatagialis caudalis (which ends only in the forearm). The cranial part along its extent lies dorsally than the caudal part. Both bellies are also present in pigeons and falcons, but in these birds the pars propatagialis cranialis (whichs inserts only to the lig. propatagialis) lies ventrally than the pars propatagialis caudalis (which ends in lig. propatagialis and in the forearm).

One tendon structure (the so‐called ‘scapular anchor’) is associated with the complex of the deltoid muscles and is present in many birds. George & Berger (1966) note that this structure extends from the dorsal surface of the scapular blade to the M. deltoideus pars major in many bird taxa; according to Canova et al. (2015) in particular in F. tinninculus it is attached to the M. deltoideus pars major. In F. sparverius, the scapular anchor is attached not only to the M. deltoideus pars major, but also to the M. scapulotriceps (Meyers, 1992). In P. erithacus, the scapular anchor is unusually attached to the M. deltoideus pars propatagialis cranialis. This shift of the attachment is obviously related to the strong reduction of the M. deltoideus pars major in parrots.

The M. flexor carpi ulnaris in P. erithacus does not have an independent small muscular belly, which is associated with the secondary feathers in a large number of bird taxa, including some passerines, pigeons (George & Berger, 1966) and also falcons: F. sparverius (Meyers, 1996) and F. tinninculus (Canova et al. 2015). The connection of this muscle with the secondary feathers in P. erithacus is by means of tendon fibers associated with the caudal aponeurosis. Distally this muscle in P. erithacus forms two independent tendons of insertion, which nevertheless insert on the os carpi ulnare side by side. These tendons are not equal: the cranial one is much larger than the caudal one. This, together with the fact that the caudal tendon of insertion is the thinner among the two, suggests that the caudal tendon of insertion likely corresponds to the additional muscle belly of the M. flexor carpi ulnaris in other birds.

The M. flexor digitorum superficialis is closely connected with the aponeurotic system of the forearm. George & Berger (1966) describe significant variability in its structure and various possible connections with the aponeurotic system of the forearm. Near its origin, the muscle is tightly bound with an aponeurotic structure called the ‘humerocarpal band’, and through this structure it is closely associated with the M. flexor carpi ulnaris in most birds (George & Berger, 1966). However, in P. erithacus, the muscle is well separated from the M. flexor carpi ulnaris. The M. flexor digitorum superficialis in P. erithacus originates from a very thick lig. humerocarpale, which is clearly distinguished from other aponeurotic structures of the forearm. In passerines and pigeons, the muscle originates on the humerus with an independent tendon of origin like in most birds (George & Berger, 1966). However, in pigeons a part of the muscle fibers originates on the humerocarpal band at some distance from the humerus, (George & Berger, 1966), demonstrating an intermediate degree of integration between the muscle and the ligament. In parrots, this muscle is thus maximally integrated with the lig. humerocarpale with simultaneous simplification of the entire non‐muscular aponeurotic system of the forearm, which is reduced to the hypertrophied humerocarpal ligament.

The general characteristics of most forearm muscles in P. erithacus is the proximal displacement of their bellies and a high extent of tendons of insertion as compared with passerines, falcons (George & Berger, 1966) and especially pigeons (Dial, 1992a). Muscular bellies of the distal wing in pigeons are well developed and often reach the distal margin of the forearm. In P. erithacus, on the contrary, the bellies are more strongly covered by aponeuroses, and much of the lengths of the corresponding muscles are represented by their tendons of insertion. This is particularly noticeable on the M. biceps brachii, M. humerotriceps, M. scapulotriceps, M. flexor digitorum superficialis, M. flexor carpi ulnaris, M. extensor carpi radialis, M. extensor carpi ulnaris and the M. extensor digitorum communis.

We also note that muscles controlling flight feathers in P. erithacus are considerably simplified, and feathers are actually only associated with the aponeurotic structures. This is manifested in P. erithacus in that the M. extensor secundariorum and an addition belly of the M. flexor carpi ulnaris are absent; this is also true for the second belly of the M. ulnometacarpalis dorsalis, which in some other birds is completely associated with the primary feathers (George & Berger, 1966). In P. erithacus the muscle has one small muscle belly and sends only a small number of fibers to the primary feathers.

Functional implications

In the last decades, experiment‐based aerodynamic models of the avian flapping flight are being progressively developed (Spedding, 1992; Spedding et al. 2003, 2009; Tobalske et al. 2003; Hedenström et al. 2006; Henningsson et al. 2008; Pennycuick, 2008; Johansson & Hedenström, 2009; Muijres et al. 2012a,b). Also, there is a number of valuable experimental studies of activity of some major flight muscles, which consider electric activity (Dial et al. 1987, 1991; Dial, 1992a,b; Meyers, 1993; Vazquez, 1995; Poore et al. 1997a; Woolley, 2000; Hedrick et al. 2003; Tobalske et al. 2003, 2010; Hedrick & Biewener, 2007; Tobalske & Biewener, 2008; Robertson & Biewener, 2012; Mahalingam & Welch, 2013), forces (Poore et al. 1997a,b; Hedrick et al. 2003; Tobalske et al. 2003) and even work loops (Hedrick et al. 2003; Tobalske et al. 2003; Tobalske & Biewener, 2008). However, these two aspects of flight (aerodynamics and muscular action) are not yet linked together in the sense of transmission of aerodynamic forces from the wing to the body. We are aware of the only attempt of this kind by Grey (1968) undertaken long ago. Grey was restricted in his analysis to very general theoretical assumptions, because at that time experimental data and reliable mathematic model of flapping flight were lacking. In the present work we attempt to interpret some of our anatomical findings in the light of current experimental knowledge of flapping flight (see references above). We discuss a number of specific features of the parrot's wing for which it is possible to propose a functional interpretation based on the most general aerodynamic suppositions.

Wing depression

The main wing depressor is the M. pectoralis. We have found in P. erithacus and P. krameri a strong connection between the tendon of insertion of the M. pectoralis and the short tendon of origin of the M. biceps brachii (Fig. 10). This feature may be universal for birds, because George & Berger (1966) mention it for Gavia, Chen, Grus and Goura. Meyers (1992) has described it in F. sparverius, and we have found a similar tendinous structure in G. glandarius, P. pica, Falco naumanni, F. vespertinus, C. garrulus and C. livia. This well‐developed structure may have an important functional role in flight, but is nevertheless very poorly discussed in the literature. Importantly, this tendinous connection anchors the M. pectoralis to the base of the crus ventrale fossae humeri, that is, on the opposite side of the humerus relative to the main insertion of this powerful muscle on the crista deltopectoralis. One functional explanation for this peculiarity (Pennycuick, 1972) implies a structural and possible functional division within the pars thoracicus of the M. pectoralis. In some birds, a well‐separated deep portion of the M. pectoralis inserts via aponeurosis just on the short tendon of origin of M. biceps brachii. This belly was found in Cathartidae (Fisher, 1946; Rosser & George, 1986), Old World vultures (Pennycuick, 1982), Ciconiiformes Gruidae (Beddard, 1898; Fisher & Goodman, 1955), Procellariiformes (Kuroda, 1961; Pennycuick, 1982) and Pelecaniformes (Beddard, 1898; Kuroda, 1961). This deep portion is represented by slow muscular fibers distinguished form the main portion by a lighter color (Rosser & George, 1986), and is thus regarded as an adaptation to soaring or gliding (Pennycuick, 1972; Meyers, 1992). However, such a separation of the M. pectoralis pars thoracicus is not universal. Psittacus erithacus does not have this separate deep portion, which also has not been found in Falco, Corvidae (Shufeldt, 1890; Meyers, 1992; personal observations), Cinclus mexicanus and Grus canadensis (Meyers, 1992). The latter author explains the additional insertion of the M. pectoralis without any structural separation of the deep portion as an intermediate state between specialized and non‐specialized gliding flight.

We argue that the above‐described additional insertion of the M. pectoralis may not be associated with any adaptation to soaring or gliding, or at least it does not represent a consequence of such a flight specialization. This attachment is present in many birds that do not use such flight types (George & Berger, 1966; see also above), and likely has a more general mechanical explanation. By means of this additional tendinous slip the M. pectoralis attaches to the side of the proximal humerus that is opposite to the main attachment. Both attachments are positioned away from the longitudinal axis of the humerus, and thus the depressing force exerted by the M. pectoralis would also rotate the humerus. In the initial phase of the wing depression, the dorsal margin of the humerus is elevated (see supplemental videos 2 and 3 in Gatesy et al. 2010) and thus the main tendon of insertion is tense. The M. pectoralis would pronate the humerus by the main tendon as long as the additional tendon is relaxed. Increased pronation during wing depression leads to the elevation of the ventral margin of the humerus relative to its dorsal margin, and thus causes tension of the additional tendon of insertion. This tension produces a supination effect on the humerus and thus stops or at least retards pronation. Consequently, this double insertion may be viewed as a mechanism that automatically controls degree of rotation of the humerus. This mechanism is certainly not a specific feature of parrots. The presence of the additional insertion of the M. pectoralis on the opposite side of the humerus in different bird groups suggests that this feature was either repeatedly acquired independently or inherited from a common ancestor. In any case, the broad taxonomic distribution of this mechanism (see above) makes it an important part of bird flight apparatus in general.

Among the muscles of the shoulder, electrical activity during downstroke is also reported for the M. coracobrachialis caudalis, M. subcoracoscapularis and M. sternocoracoideus (Dial et al. 1991). It is obvious that none of the listed muscles can compete with the M. pectoralis in wing depression, as their combined physiological cross‐sectional area, and hence the available force, is at least five times smaller than that of the M. pectoralis (our data for fixed and fresh specimens of P. erithacus). However, they can help and correct the effect produced by the M. pectoralis in certain special respects. Two of the above‐mentioned muscles have specific features in parrots. The caput ventrale of the M. subcoracoscapularis in P. erithacus originates from the membrana sternocoracoclavicularis rather than the coracoid as it usually does in other birds (George & Berger, 1966). In addition, the belly of the muscle is not pennate but consists of long parallel fibers. Typically, the muscle originates from the shaft of the coracoid, for example in G. glandarius. Due to the different origin and fiber architecture, the calculated ratio of the length of the muscle fibers to the length of the coracoid in G. glandarius is 0.42, while in P. erithacus it is 0.56. The enlarged distance between the points of origin and insertion of the muscle leaves a space for the long‐fibered muscle belly, which provides a larger amplitude of contraction. The long fibers of the muscle imply increased range of axial rotation of the humerus at the shoulder joint as the muscle pronates the humerus.

M. sternocoracoideus in P. erithacus is relatively large and it originates extensively on the sternum and ribs, whereas usually in birds it originates on the sternum only (George & Berger, 1966). During contraction, the muscle pulls coracoid laterally. The overall effect of this muscle largely depends on the structure of the sternocoracoid joint. In P. erithacus, the latero‐medial sliding of the coracoid along the sternum is impossible, as confirmed by in situ manipulations and also by the equality of the lengths of the articular surfaces of the coracoid and sternum. The coracoid is fixed against the sternum by two short ligaments (lig. sternocoracoideum internum, lig. sternocoracoideum externum). As a result, the main movement of the coracoid can be characterized as rocking: the scapular end of the coracoid deviates slightly medially (Fig. 25a) or laterally (Fig. 25c) from the neutral position (Fig. 25b), when the whole articular surface is in contact with the sternum. Such a rocking motion has small amplitude (about 10°) as the ligaments fixing the coracoid are short, but it is still sufficient for the distance between the scapular ends of the coracoids to vary markedly. The muscle fibers of the M. sternocoracoideus are oriented almost parallel to the shaft of the coracoid; laterally offset processus lateralis provides the muscle with a good lever arm around the sterno‐coracoid joint especially when the coracoid is declined medially from the neutral position (Fig. 25a). The M. sternocoracoideus acquires an increased lever arm to return the coracoid from the medial to the neutral position, because when the bone is in its medial position the rotational axis of the coracoid is shifted to the medial extremity of its sternal articular surface, i.e. at the maximal possible distance away from the line of action of the M. sternocoracoideus. Thus, it can be concluded that the M. sternocoracoideus in P. erithacus most likely very effectively returns the coracoid from the medial to the neutral position due to both the rocking mobility of the sternocoracoid joint providing the shift of its axis of rotation, and the lateral expansion of the muscle onto ribs.

Figure 25.

Ventral view of the shoulder girdle of the African grey parrot showing coracoid mobility and lever arms of the operating forces. (a) Coracoids tilted medially, rotational axis shifted medially, lever arm of the M. sternocoracoideus is maximal; (b) neutral position, in which the articular surfaces of the coracoids and sternum are in full contact; (c) coracoids tilted laterally, rotational axis shifted laterally, lever arm of the clavicular force is maximal while the lever arm of the M. sternocoracoideus is minimal. Abbreviations: F, furcula; C, coracoid; S, sternum; R, rib; insSC, the area of insertion of the M. sternocoracoideus on the dorsal side of the coracoid; circles indicate the axis of coracoid rocking around the sternum.

The M. sternocoracoideus is electrically active during the last third of the upstroke, the upstroke–downstroke transition and the beginning of the downstroke (Dial et al. 1991). It can thus be assumed that the maximum force of this muscle occurs in the upstroke–downstroke transition and the first third of the downstroke, taking into account a delay between the electric excitation and the development of the contraction force (Hedrick et al. 2003). At this time interval, the furcula is just beginning to widen from the most compressed position (data of Jenkins et al. 1988 on Sturnus vulgaris). If so, the lever arm of the M. sternocoracoideus is the best at this moment (our data on P. erithacus). Between the compressed and the neutral positions, the elastic recoil of the furcula helps the left and right mm. sternocoracoidei to return coracoids to the neutral position, although the lever arm of the force exerted by the furcula relative to the sternocoracoid joint is reduced in this case (Fig. 25a).

During the late downstroke, the coracoids are spread beyond the neutral position (Jenkins et al. 1988). This is caused by the forcible contraction of the M. pectoralis as is proved by its electric stimulation (Zimmer, 1935). Being expanded together with the coracoids, the furcula begins to act in the opposite direction (back again towards the neutral position) and so becomes an antagonist to the both Mm. sternocoracoidei (Fig. 25c). Note that the axis of rocking of the coracoid on the sternum is shifted here to the lateral extremity of the sternocoracoid joint. Thus, the lever arm of the force of the clavicle is maximum here, while the lever arm of the M. sternocoracoideus is minimal. Not surprisingly, the decrease of mechanical efficiency of force application for this muscle coincides with the muscle relaxation: its activity is known to vanish by the mid‐downstroke (Dial et al. 1991).

As was stated above, the M. sternocoracoideus is active and most effective in respect of the lever arm at the upstroke–downstroke transition, when the wing is maximally raised and the coracoids with furcula are maximally compressed. Obviously, the M. sternocoracoideus activity is aimed at counteracting some other force that causes this compression. But what is the source of this force? Jenkins et al. (1988) showed an unexpected (and counter‐intuitive) compression of the furcular rami during the activity of the M. pectoralis – an effect that remained unexplained by the authors and contradicts the opposite observation of Zimmer (1935). Probably the compressive/expansive effect of M. pectoralis on the furcula may vary depending on the position of the humerus (i.e. the stage of the stroke) due to different geometry of force applications. We suppose that the force produced by the M. pectoralis together with the aerodynamic lift would ‘press’ the humerus into the glenoid when the wings are at their highest position, and thus may explain the observed medial deflection of the coracoids in the upper position of the wing during the downstroke initiation. Indeed, electromyographic research shows an electrical activity of the M. pectoralis at this point of the wingbeat cycle (Dial et al. 1991; Hedrick et al. 2003). The M. sternocoracoideus activates somewhat earlier than the M. pectoralis (Dial et al. 1991). Activation in advance may help this muscle to withstand the compressive effect of much stronger antagonist, and return the coracoids in the neutral position, when the action of the pectoralis changes from compressive to expansive later in the downstroke.

Wing elevation

The most significant peculiarities of the forelimb musculature in P. erithacus and parrots in general are those that are associated with wing elevation. Wing elevators are the complex of deltoid muscles and the M. supracoracoideus, which are discussed in detail above. Here we address a possible functional significance of the observed muscular arrangement.

The most unusual feature of the parrot wing morphology is the very large M. supracoracoideus (see above). The role of the hypertrophied M. supracoracoideus for parrot flight never drew any attention, to our knowledge. Only M. pectoralis was experimentally studied in parrots (Hedrick et al. 2003). In order to approach understanding of this hypertrophy, it is necessary to analyze: (1) the effect of the M. supracoracoideus on the humerus; and (2) the contribution of this muscle to the total mass of the flight musculature. Traditionally, the M. supracoracoideus is regarded as the wing elevator (Dial, 1992a). However, studies by Poore et al. (1997a,b) have shown that in starling the primary role of this muscle is to impart a high‐velocity rotation (supinaton) of the humerus about its longitudinal axis, which rapidly elevates the distal wing. This positions the wing for the initiation of the next downstroke. This implies wing folding to avoid interaction with the air. Indeed, Poore et al. (1997b) have shown that the rotational action of the M. supracoracoideus depends on the degree of the retraction of the humerus: it comes to nothing when the humerus is protracted. In situ joint manipulation in Cairina moschata has supported (our data) that the M. supracoracoideus applies supination torque to the humerus only when the humerus is in the retracted position (adducted to the body); in the more protracted position of the humerus, the rotation torque of the M. supracoracoideus is indeed slightly pronational. However, simply positioning the half‐folded wing for the next downstroke does not require a great muscular force if this action occurs with minimal interaction with the air. Such upstroke may be called ‘passive’ in contrast to the ‘active’ upstroke ensuring aerodynamic force generation.

The hypothesis of high‐velocity supination of the wing as preparation to the downstroke without aerodynamic force generation (passive upstroke) is obviously not enough to explain the hypertrophy of the M. supracoracoideus in parrots. The weight of the M. supracoracoideus in parrots constitutes about 25% of the mass of the M. pectoralis (our data for fixed and fresh specimens of the P. erithacus and P. krameri), whereas in other birds the proportion of the M. supracoracoideus ranges from 5% to 10% of M. pectoralis (our data for the Dryocopus martius, Dendrocopos major, F. vespertinus, F. tinnunculus, G. glandarius). In the Anseriformes, Rallidae and Charadrii, the proportion of M. supracoracoideus varies between 10% and 15% (Sych, 1999). Such a significant difference indicates that the M. supracoracoideus in parrots performs mechanical work comparable (though smaller) to that of M. pectoralis (Biewener, 2011). Such work in flight is only possible when some significant aerodynamic force (lift plus wing thrust) is produced. The most useful thrust in steady flight is the one that is directed upwards and slightly forward (against body drag). This best aerodynamic force can be produced only during the downstroke. However, it is possible to produce an upward and backward directed aerodynamic force during the upstroke (Hedenström, 2002). Gravity force is the main problem during aerial flight, and thus any upward directed force would be useful even if it is associated with some retarding byproduct. However, in order to transfer the useful upward component of the aerodynamic force to the body, a muscle should pull the body up and hence the wing down (according to Newton's third law). But none of the wing elevators can pull the wing down and the body up when the body is oriented horizontally.

However, muscle‐elevators of the wing can produce upward aerodynamic force during upstroke when the body is vertically oriented. One very well‐studied example is the hovering flight of hummingbirds, which have extremely large M. supracoracoideus with a mass that varies from 40% to 70% of the mass of the M. pectoralis (Altshuler et al. 2010). In hovering hummingbirds, the supracoracoideus muscle provides an active upstroke, with the upward momentum constituting about 25–33% of the total per wing‐beat cycle (Warrick et al. 2012). The ability to produce the upward aerodynamic force during the upstroke with the vertically oriented body (during flight maneuvers) is also experimentally shown for pigeons: their upstroke aerodynamic force is up to 50% of the aerodynamic force produced during the downstroke (Ros et al. 2011). Accordingly, the M. supracoracoideus in pigeons is large (14% of the weight of the M. pectoralis in Columba palumbus; Sych, 1999). During this kind of upstroke, the ventral surface of the wing faces upwards and meets the air with a negative incidence (angle of attack). The resulting aerodynamic force is thus directed forward and upward (cranioventally in the body coordinates). The M. supracoracoideus mainly applies supination torque to the humerus and pulls the wing dorsally and caudally, and pulls the body cranioventrally. Thus, the action of the M. supracoracoideus can produce during the upstroke a useful aerodynamic force during flight with a vertically oriented body. Another example of such a flight is the vertical take‐off typical to galliforms. Remarkably, the ratio of the wing elevators in galliforms is quite similar to that of parrots. In various galliforms, the mass of the M. supracoracoideus varies from 22% to 34% of the M. pectoralis, and the M. deltoideus pars major varies from 0.5% to 2% of the pectoralis4 (Sych, 1999; Yang et al. 2015). The great relative development of the M. supracoracoideus in parrots (which in fact is more comparable to this muscle in galliforms) leads us to the conclusion that parrots also may actively use such a flight.

The forceful thrust by the supracoracoideus muscle during the upstroke, which we presume for low‐speeed flight with vertical or inclined body, is well supported by experimental data. According to Tobalske & Biewener (2008), in the ascending flight of pigeon the power produced by the supracoracoideus muscle during the upstroke is equal to about 40% of the power produced by the pectoralis muscle during the downstroke. In the level or descending flight, the respective ratio of the supracoracoideus power during the upstroke to the pectoralis power during the downstroke is a little lower, about 30%. The efficiency of the supracoracoideus muscle in producing an active thrust against the air is ensured by inclined body position in both ascending and descending flight. The same is true for the slow level flight; in the experiments of Tobalske & Biewener (2008) the pigeon's horizontal velocity was no more than about 4 m s⁻¹, because the flight distance from perch to perch was small, and the pigeon had no opportunity to fly faster. Kinematic data provided by Berg & Biewener (2010) show that, at such low speeds of flight, pigeons keep the body in the inclined position indeed. So, all the measurements of the supracoracoideus power production, which we rely upon, come from the flight with a non‐horizontal body. In our opinion, this is the reason for such a high power production of the supracoracoideus muscle, which was not expected by the authors of the experiments. Both Poore et al. (1997a) and Biewener (2011) attempt to explain the extra supracoracoideus power by the wing inertia, which should be overcome at the downstroke–upstroke transition when the wing is near its lowermost point. In our opinion, this action cannot require the supracoracoideus power as high as 30% of the pectoralis power during the downstroke. In fact, to overcome the wing inertia at its lowermost point and reverse its movement relative to the body, it is enough to relax the pectoralis muscle and allow the body to fall passively from the wings while they are still supported by air resistance. The huge supracoracoideus power is better explained by its involvement in propagation of the useful aerodynamic force from the wing to the body during the upstroke, similar to what is observed in the hovering hummingbird. The non‐horizontal body position matters in both cases.

The proposed hypothesis of a possible active upstroke during flight with vertically (non‐horizontally) oriented body in parrots is in accordance with the reduction of the M. deltoideus pars major, which is useless and even harmful for vertical flight. This muscle pulls the humerus dorsally and the body ventrally, thus increasing the unnecessary horizontal (anatomically ventral) body oscillations. It also protracts the humerus and pulls the body caudally, which for vertical flight means downward. The M. deltoideus pars major is not needed when the M. supracoracoideus is hypertrophied, because the latter muscle is powerful enough to elevate the wing. The combined mass of M. supracoracoideus and M. deltoideus pars major of the studied non‐psittaciform birds ranged from 10% to 16% of the mass of M. pectoralis, and only in D. martius it slightly exceeded 20%. But in parrots, the mass of the M. supracoracoideus alone significantly exceeds these values, and thus the evolutionary reduction of the M. deltoideus pars major is reasonable.

The proposed hypothesis can further explain the increase of the M. deltoideus pars propatagialis in parrots. The mass of this muscle (pars cranialis + pars caudalis) is more than 4% of the pectoralis mass in the grey parrot and is more than 5% in the rose‐ringed parakeet (our data). The combined mass of the propatagial portions did not exceed 3% of the pectoralis mass in the other studied birds. In particular, parrots have a very large M. deltoideus pars propatagalis cranialis (see above). This portion applies a considerable supination torque to the entire wing during the upstroke (it supinates the wing from the pronated position) as it is attached to the ventral side of the wrist via the ligamentum propatagialis. The m. deltoideus pars propatagalis cranialis may be very useful during the vertical take‐off because it assists in positioning the wing at the preparatory phase before the upstroke with a negative angle of attack (the M. supracoracoideus supinates the base of the wing, but it is necessary to supinate the distal part of the wing even more to properly adjust the angle of attack by the primaries).

The scapular anchor is the tendinous structure associated with the deltoid complex. Meyers (1992) suggested that it can limit the protraction of the humerus or regulate the position of the humerus in the shoulder joint. He also supposed that it may have a function in the gliding flight despite the fact that this structure occurs in many birds that do not use such flight. We assume that in parrots this structure shifts the area of origin of the deltoid muscle caudally by linking the largest (propatagial) part of the deltoid complex to the middle of the scapula. This shifts the line of action of the deltoid muscles posterior to the glenoid and so makes them retractors rather than protractors (Fig. 26). This effect also conforms to the proposed hypothesis because during vertical flight the upstroke with the negative angle of attack necessitates retractive, not protractive, muscular forces applied to the wing.

Figure 26.

Dorsal view of the wing of the African grey parrot showing the effect of the scapular anchor on the resulting force of the M. deltoideus pars propatagialis cranialis. (a) A hypothetical average force direction of the M. deltoideus pars propatagialis cranialis whithout the scapular anchor; (b) the inferred actual average force direction of the M. deltoideus pars propatagialis cranialis in the parrot's wing with the scapular anchor. Abbreviations: Fdpl, the average force direction of the M. deltoideus pars propatagialis cranialis; ScAn, scapular anchor.

All these functional interpretations of the anatomical observations lead us to the conclusion that flight with the vertically oriented body is not only possible but may play a significant role in parrot locomotion. The grey parrot may use vertical flight in the forests where they nest in hollows and feed on both the tree crown and on the ground (Athan & Deter, 2000). The vertical flight can be useful for all forest‐dwelling parrots for a quick rise from the ground or lower branches into the crowns in the dense tropical vegetation. The fact that observed anatomical observations are likely common to all parrots (see above) supports an idea that adaptation to vertical flight may be one of the common features of the Psittaciformes. The ability of vertical flight and the zygodactyl foot together link the origin of parrots with the dense (likely tropical) forests.

Author contributions

N.Z., A.P. and D.R. conceived the study. D.R. performed the dissections, collected data, prepared all figures and drafted the manuscript. D.R., A.P. and N.Z. analyzed the data and wrote the manuscript. A.P. provided all studied fresh specimens, N.Z. provided studied skeletons.

Supporting information

Table S1 Masses of the main pectoral girdle and forelimb muscles of the African grey parrot (Psittacus erithacus).