Kinematics of Chiropteran Shoulder Girdle in Flight


Correspondence to: A.A. Panyutina, Department of Vertebrate Zoology, Biological Faculty of Moscow State University, Moscow 119234, Russia. E-mail:


New data on the mechanisms of movements of the shoulder girdle and humerus of bats are described; potential mobility is compared to the movements actually used in flight. The study was performed on the basis of morphological and functional analysis of anatomical specimens of 15 species, high speed and high definition filming of two species and X-ray survey of Rousettus aegyptiacus flight. Our observations indicate that any excursions of the shoulder girdle in bats have relatively small input in the wing amplitude. Shoulder girdle movements resemble kinematics of a crank mechanism: clavicle plays the role of crank, and scapula—the role of connecting rod. Previously described osseous “locking mechanisms” in shoulder joint of advanced bats do not affect the movements, actually used in flight. The wing beats in bats are performed predominantly by movements of humerus relative to shoulder girdle, although these movements occupy the caudal-most sector of available shoulder mobility. Anat Rec, 296:382–394, 2013. © 2013 Wiley Periodicals, Inc.

In the second half of the 20th century many studies on osteology and myology of bats were published (Vaughan, 1959, 1970; Norberg, 1970, 1972; Kovtun, 1976-1978). In addition to descriptive anatomy, attention was paid to comparative analysis of morphological adaptations of different members of the order (Strickler, 1978; Schlosser-Sturm and Schliemann, 1995). External wing shape was associated with flight style and habitat use (Norberg and Rayner, 1987; Kruskop, 1999; Ševčik, 2003; Patriquin and Barclay, 2003).

Movements of the distal wing segments in bats are readily accessible to observation and are rather thoroughly investigated (Tian at al., 2006; Swartz et al., 2006). On the contrary, movements of elements of the shoulder girdle are hidden in the body and are difficult to study. Experimental results considering the action of shoulder girdle in bats during flight are limited to a few articles on electromyography (Kovtun and Moroz, 1973, 1974; Hermanson and Altenbach, 1983, 1985; Altenbach and Hermanson, 1987), in the latter of which the kinematics of humerus and scapula are also discussed on the basis of surveys of animals with steel pins implanted in these bones, and one on radiography (Hermanson, 1981). The X-rays were taken on animals which were able to move the wings freely being grasped by the uropatagium. In addition, these X-rays were single-frame exposures, and no X-ray filming of a flying bat was ever published. In the article cited, Hermanson provides only two radiograms of Antrozous pallidus in the frontal plane, which he judged to correspond to normal flight patterns, despite the asymmetric position of the wings. Hermanson did not consider the movements of scapula, and for the clavicle he described only latero-medial deflections in the transverse plane (movements that he called adduction and abduction). Although Hermanson had no opportunity to evaluate them quantitatively, he concluded that the clavicle moves in phase with the wing: when the wing goes down, the clavicle deflects with it laterally, and during the upstroke—medially.

In the absence of reliable data on the position of scapula and clavicle at different stages of wing beat cycle, it is impossible to correctly interpret the results of electromyography or to analyze the load distribution on the muscles, bones and ligaments of shoulder girdle.

Preliminary results of our study of scapular and clavicular principal movements in Rousettus aegyptiacus (R. aegyptiacus), based on fluoroscopy, were published elsewhere (Panyutina et al., 2011). In this article, results of fluoroscopy are considered in view of shoulder movements in different bat species revealed by high-speed video and examination of syndesmological specimens.



We studied the structure of the forelimb articulations in 11 bat species belonging to five families (Table 1, column 1). Joint mobility was studied on syndesmological preparations of seven species belonging to five families and on fresh corpses of eight species of four families (Table 1, column 2).

Table 1. Materials
SpeciesOsteological studyStudy of mobility (qualitative analysis)Study of mobility (quantitative analysis)Video recording experimentsX-ray experiments
Cynopterus sphinx VahlSkeleton, alcohol syndesmological preparationAlcohol syndesmological preparation
Rousettus aegyptiacus GeoffroySkeleton, alcohol syndesmological preparationFresh corpse, alcohol syndesmological preparation, fresh syndesmological preparationFresh syndesmological preparationLive animal
Rousettus leschenaultii DesmarestLive animal
Pteropus lylei AndersenSkeleton
Pteropus tonganus Quoy et GaimardSkeleton
Rhinolophus ferrumequinum SchreberSkeleton, alcohol syndesmological preparationAlcohol syndesmological preparation
Rhinolophus borneensis PetersSkeleton
Hipposideros larvatus HorsfieldAlcohol syndesmological preparationAlcohol syndesmological preparation
Hipposideros armiger HodgsonSkeleton, alcohol syndesmological preparationFresh corpse, alcohol syndesmological preparationAlcohol syndesmological preparationLive animal
Myotis blythi TomesSkeleton
Myotis daubentonii KuhlFresh corpse
Myotis dasycneme BoieFresh corpse
Plecotus auritus LFresh corpse
Eptesicus nilssoni Keyserling et BlasiusFresh corpse
Nyctalus noctula SchreberFresh syndesmological preparationFresh corpse, fresh syndesmological preparation
Carollia perspicillata LAlcohol syndesmological preparationFresh corpse, alcohol syndesmological preparation, fresh syndesmological preparation

To quantify the limits of the shoulder joint mobility, syndesmological preparations of freshly dissected R.aegyptiacus as well as fixed Hipposideros armiger were used. Fixed specimens were placed in 80% ethanol for one month (Table 1, column 3). The preparations included the scapula and humerus in a natural articulation with ligaments and parts of supraspinatus, infraspinatus, subscapularis and triceps muscles adjacent to the capsule of shoulder joint. During the experiment, the scapula was fixed on a metal plate with CA glue, and the plate was gripped in a vise. The humerus was moved by means of reins of thin fishing line tied to its distal part; the reins were supplied with elastic rubber bands at their ends for gentle application of pulling forces produced by experimenter's hands. The humerus was turned in the same direction in which it moves in flight. The full range of available circumduction of humerus relative to scapula was rendered. Successive stages of humerus circumduction were photographed with Nikon D300S camera with 60 mm f/2.8D AF Micro-Nikkor lens in lateral, dorsal, and anterior views. The details of experiment can be seen in Movie S1 in Supporting Information.

X-Ray Experiments

The mobility of the shoulder girdle elements was studied using fluoroscopy of R. aegyptiacus flight. Filming was performed with an X-ray system Philips Allura Xper FD 20 at 30 shots per second, frame size 880*880 pixels. More than 5,000 shots of flying animals were obtained. For filming a collar with a leash was put on an animal's neck. The leash was held taut to keep the bat in the frame (Fig. 1). Staying in the frame, the bat's body had a very low speed relative to the air, and so, to keep itself airborne, the animal was obliged to beat its wings more forcibly than in normal progressive flight.

Figure 1.

Scheme of X-ray recording experiment.

Video Recording Experiments

To assess the adequacy of locomotor cycle representation in the X-ray records (30 fps) the high speed (300 fps) videography of two species was conducted: R. leschenaultii (species close to R. aegyptiacus) and H. armiger (rhinolophoid bat with highly specialized skeletal features of thorax and shoulder joint). In these experiments, the bats were not constrained by any leash or collar, but were flying freely in the rather spacious enclosure (height—170 cm, length—200 cm, and width—110 cm) with roosts. Filming was performed by Casio EX-F1 camera (300 fps, shutter speed 1/4,000 s, frame size 512*384 pixels). Simultaneously, to obtain high quality images of key stages of the wing beat cycle high definition shooting was carried out by Panasinic HDC-SDT750 (50 fps, shutter speed 1/8,000 s, frame size 1,920*1,080 pixels).

All the experiments with live animals were performed under control of the Moscow State University Bioethics Committee.


Available Mobility of Shoulder Girdle and Humerus

Sternoclavicular joint

Articular surfaces of the sternum and clavicle are rather flat and oval-shaped. In addition to the thin-walled joint capsule, there are two well-developed ligaments in this joint, significantly limiting the mobility of the clavicle. One of them is the obliquely oriented medial sternoclavicular ligament (ligamentum sternoclaviculare mediale), which connects the medial border of the proximal head of clavicle with the front edge of manubrium sterni (Fig. 2A). The second one, that is the lateral ligament (ligamentum sternoclaviculare laterale), is shorter and located on the lateral side of the joint (Fig. 2B). It connects the lateral edges of the articular surfaces of the sternum and clavicle immediately adjacent to each other. The location of these short ligaments significantly restricts disarticulation of the flattened articular surfaces, leaving, however, some mobility (Fig. 2).

Figure 2.

Degrees of freedom of the shoulder girdle in bats, anterior view. Position (A): the clavicle is in its medial position; dotted line shows limits of the clavicle mobility; arrows indicate the direction of movement toward the next position. Position (B): the clavicle abducted and rotated; previous position is shown in gray

(Modified from Panyutina et al., 2011)


The main movement in the joint is rotation of the clavicle about the longitudinal axis of its proximal part (the stem). The limits of rotation are determined by the length of the medial ligament, which is winding around the base of the clavicle with its turning. It should be noted that the chiropteran clavicle is more or less sharply bent in its distal part (Fig. 3), and hence, the rotation about its proximal part causes arc-like movements of its acromial end respective to thorax (Fig. 2B). The range of potentially available clavicle rotation varies somewhat in various groups of bats: in more primitive forms the mobility is greater than in the specialized bats, such as members of Hipposideros genus, bearing on the dorsolateral surface of the base of clavicle a flat prominence (indicated by arrow in Fig. 3), which forms an additional strut against manubrium sterni. In any case, the axial rotation of clavicle exceeds its latero-medial mobility, which in its turn exceeds the cranio-caudal one.

Figure 3.

Right clavicle of H. armiger, lateral view. Dashed line—longitudinal axis of clavicle as a whole, solid line—longitudinal axis of its proximal part (the stem). Arrow points on the additional tubercle at the base of the clavicle.

Acromioclavicular joint

In addition to the mobility of the clavicle relative to the axial skeleton, movements between the clavicle and scapula also occur in bats. This mobility is limited by two ligaments. The first one, ligamentum acromioclaviculare (Fig. 2A) is a strong band between the ends of acromion and clavicle and is often strengthened by the sesamoid. The second, ligamentum coracoclaviculare (Fig. 2B) connects the cranial edge of the base of coracoid process and caudolateral surface of acromial end of clavicle. Thus, the distal end of the clavicle is tied up by these ligaments between the acromial and coracoid processes at the cranial end of scapula. Together these ligaments allow a fairly wide range of mobility, however, the main implemented movement is the rotation of scapula about its longitudinal axis (Fig. 2B), passing through the acromioclavicular joint and the dorsocaudal angle of scapula. Meanwhile, ligaments wind around the distal end of clavicle, delimiting by their length the range of this rotation. In some bats, such as members of Rhinolophidae and Hipposideridae, the coracoclavicular ligament is much shorter than the acromioclavicular one, which means that the end of clavicle is tightly bound to the base of coracoid process. However, the relative lengths of the ligaments do not significantly influence the mode of the scapular mobility relative to the clavicle in various bats.

Shoulder joint

Previous detailed studies of the shoulder joint in bats (Schlosser-Sturm and Schliemann, 1995) revealed three main types of its structure: the basic nonspecialized joint, the specialized joint with one articular surface and the specialized joint with two articular surfaces. In the basic type of the joint (e.g., in Pteropodidae, Rhinopomatidae, and Nycteridae) the head of humerus is spherical, and the glenoid cavity of scapula is, respectively, nearly round. In the first type of specialized joint (e.g., in Noctilionidae, Mormoopidae, Emballonuridae, and Megadermatidae) the head of humerus and the glenoid cavity are noticeably elongated in cranio-caudal direction. The second type of specialized joint (e.g., in Vespertilionoidea, Rhinolophidae, Craseonycteridae, and Phyllostomidae) is characterized by an additional articulation between the humerus and scapula. It is formed by the hypertrophy of the greater tuberosity of humerus, which rests against corresponding articular facet on scapula located above the glenoid cavity. In some species, in addition, supraglenoid tubercle of scapula is enlarged and rests between greater and lesser tuberosities of humerus when the latter is extremely protracted (Vaughan, 1959).

In all studied bats, the joint capsule has no pronounced ligaments and its walls are strengthened by tendons of adjacent muscles: supraspinatus, infraspinatus, subscapularis, and caput longum of triceps brachii. A wide range of humerus mobility in the shoulder joint is possible: abduction/adduction in the transverse plane, retraction/protraction in the frontal plane, and rotation around the longitudinal axis of the humerus. Abduction/adduction, combined with retraction/protraction give circumduction, which limits are revealed by our study of syndesmological preparations (Figs. 4-6). In both studied species humerus abduction exceeds adduction, that is why the elbow trajectory in the side view (Fig. 4) is eccentric with respect to the shoulder joint. Comparison of humerus mobility in syndesmological preparations of Rousettus (the basic type of shoulder joint according to Schlosser-Sturm and Schliemann, 1995) and Hipposideros (specialized joint of the second type of Schlosser-Sturm and Schliemann) shows almost identical available limits of the shoulder joint mobility. The most pronounced differences exist in caudoventral sector of the trajectory: when the humerus is retracted, Rousettus is able to adduct (depress) it slightly further than Hipposideros (Figs. 4C,5C,6C).

Figure 4.

Range of humerus mobility relative to scapula on syndesmological preparations, right side view, the planes of scapulae oriented horizontally. (A) R. aegyptiacus, fresh specimen. (B) H. armiger, fixed in alcohol. (C) Both elbow trajectories together scaled to equal humerus length and superimposed by the center of the shoulder joint.

Figure 5.

Range of humerus mobility relative to scapula on syndesmological preparations, right side in dorsal view, longitudinal axes of scapulae oriented horizontally, the planes of scapulae in the picture plane. (A) R. aegyptiacus, fresh specimen. (B) H. armiger, fixed in alcohol. (C) Both elbow trajectories together scaled to equal humerus length and superimposed by the center of the shoulder joint.

Figure 6.

Range of humerus mobility relative to scapula on syndesmological preparations, right side in anterior view (along the longitudinal axis of scapula), the planes of scapulae oriented horizontally. (A) R. aegyptiacus, fresh specimen. (B) H. armiger, fixed in alcohol. (C) Both elbow trajectories together scaled to equal humerus length and superimposed by the center of the shoulder joint.

Actual Movements of Shoulder Girdle in Flight

The study of skeletons and syndesmological preparations is essential but not sufficient in determining the ongoing movement of shoulder girdle, as bats do not use the whole range of available mobility. Actual movement of the shoulder girdle in flight we have obtained from X-ray records (see Movie S2 in Supporting Information). In spite of the collar and leash, the wing movements while X-ray survey were rather natural as evidenced by high speed video records of the free flight (see Movies S3, S4 in Supporting Information). On the basis of high speed video those sequences of X-ray film were selected for analysis that covered the whole wing beat amplitude.

Movements of the shoulder girdle in flight represent repeated cycles of the same duration as the wing beat cycles. During the second half of upstroke and the first half of downstroke, the clavicle deflects laterally (abduction) (Fig. 7A,B,A′). At the transition from upstroke to downstroke its rotation in the direction of supination starts, which increases the lateral deviation of acromion process due to the distinctive curvature of the clavicle (see above). At the middle of downstroke, acromion comes to its lateral-most point, and then the medial deviation (adduction) of the clavicle begins (Fig. 7B,C). At the transition from downstroke to upstroke the clavicle begins to deflect cranially (protraction), and its supination changes to pronation. When the acromion reaches the anterior-most point of its trajectory (Fig. 7D) the clavicle starts its caudal deflection (retraction) (Fig. 7E). At the middle point of upstroke adduction of the clavicle changes back to abduction and the cycle repeats. Characteristically, the cranio-caudal shifts of clavicle (protraction and retraction) are confined to the wing upstroke, but throughout the most loaded phase of the cycle, the downstroke, it stays caudally deflected. Phase relations of the three above types of clavicle movements are shown in Figure 8A.

Figure 7.

X-ray images of successive wing beat phases of R. aegyptiacus. Sagittal and frontal views were shot separately. Here, wing beat cycle duration was 6/30 s and 5/30 s for the sagittal and frontal views, respectively. To bring stages in both series into better accordance one frame between c and d in sagittal view was omitted on this diagram (see the full sequence in Movie S2 in Supporting Information). (A and A′) Wings are at their upper position (point of transition from upstroke to downstroke). (B and C) Downstroke. (D and E) Upstroke. On the frontal view on one side of the body, the clavicle is filled with blue, and the scapula is outlined by red. 1—sternoclavicular joint; 2—full trajectory of the distal end of clavicle; 3—acromioclavicular joint; 4—dorsocaudal angle of scapula; 5—full trajectory of the dorsocaudal angle of scapula; 6—direction of the proximal part (the stem) of clavicle; 7—the limits of the clavicle stem movements in frontal plane; 8—degree of protraction/retraction of the distal end of clavicle; 9—the limits of protraction/retraction of the distal end of clavicle.

Figure 8.

Basic scheme of the complete cycle of the shoulder girdle mobility of R. aegyptiacus in flight (based on X-ray data). (A) Phase relations of the three types of clavicular movements. (B) Phase relations of the same movements, superimposed on the trajectory of acromion relative to thorax. Frontal view, one half of the body is shown. Arrows indicate the direction of movements. 1—midline of the body; 2—trajectory of the dorsocaudal angle of scapula; 3—sternoclavicular joint; 4—trajectory of the distal end of clavicle; 5—extreme medial position of acromioclavicular joint; 6—phases of cranio-caudal deflections of clavicle; 7—phases of latero-medial deflections of clavicle; 8—phases of axial rotation of clavicle. (Modified from Panyutina et al., 2011).

The limits of clavicle movements in flight are drawn on X-ray images (Fig. 7): I – maximum lateral abduction, II—maximum medial adduction, i—maximum retraction, clavicle up, ii—maximal protraction, clavicle down. The amplitude of the axial clavicle rotation (supination/pronation) in R. aegyptiacus during flight is at least 45 degrees, the amplitude of medio-lateral movements of its stem does not exceed 30 degrees, and the cranio-caudal deviations are too restricted to be measured reliably.

Throughout the wing beat cycle, the scapular blade stays on the dorsal surface of thorax and does not show any significant deviations from the frontal plane. In this plane the caudal end (i.e., dorsocaudal angle) of scapula shifts anteroposteriorly along the rectilinear trajectory, which is almost parallel to the midline of the body. The trajectory of the acromial process of scapula (together with the distal end of clavicle) is sharply different from that of its caudal end. This trajectory approaches an ellipse (Fig. 8B), which larger diameter is oriented rather perpendicular to the midline, but with some declination—lateral pole of the ellipse is located slightly caudal to the medial one. Acromion passes the ellipse in the counterclockwise direction, if we look at the bat's right side from above (Fig. 8B). Craniomedial half of the ellipse is engaged in the upstroke and caudolateral one—in the downstroke. Thus, in the frontal plane, the caudal end of scapula moves rectilinearly parallel to the midline but its acromial end, driven by the clavicle, makes circular motion. Both of them reach the cranial-most points of their trajectories in the first half of upstroke, and the caudal-most points in the first half of downstroke. To describe the movements of the shoulder girdle elements, one can draw an analogy with a crank mechanism: clavicle plays the role of crank, and scapula—the role of connecting rod (here we imply the kinematical pattern only, but not the driving role of the clavicle or scapula).


Our results allow revising the concept of the chiropteran shoulder girdle mobility based on previous experiments (Hermanson, 1981). For the clavicle, besides latero-medial deflections noted by Hermanson, we also found cranio-caudal deflections and rotation around the longitudinal axis of its straight proximal part (Fig. 2). Moreover, contrary to Hermanson's conclusion, latero-medial deflections are out of phase with the wing movements but go ahead of them: the clavicle abduction begins much earlier than the downstroke. Instead, the rotation of the clavicle is almost in phase with the wing beats, and its amplitude exceeds the range of adduction/abduction. What Hermanson regarded as clavicular adduction/abduction appears to be a mixture of pure adduction/abduction and rotation.

Scapular movements are described here for the first time. Until now conceptions on this subject were based on the idea of the existence of locking mechanisms between the humerus and scapula in some groups of bats such as the families Hipposideridae, Rhinolophidae, Vespertilionidae and Phyllostomidae which representatives we have studied by various means except X-ray filming.

For the first time, the significance of joint locks for bat wing action was suggested by Vaughan (1959, 1970). He presumed that at the highest point of upstroke the greater tuberosity of humerus locks against the additional articular facet of scapula above the glenoid (specialized shoulder joint of the second type according to Schlosser-Sturm and Schliemann, 1995), which locking promotes the involvement of strong serratus ventralis muscle, connecting the ribs with the caudal edge of scapula (in chiropterans usually termed the lateral edge), into the braking of upstroke. Vaughan argued that in case of high wing beat frequency such a mechanism stopping upstroke would provide recover time for pectoralis muscle, because the beginning of the downstroke could be performed by serratus ventralis muscle if its force applied to scapula was transferred onto humerus via the proposed lock. Subsequently, this idea has not been confirmed, as it was shown that at the downstroke beginning the pectoralis muscle starts electric activity earlier than the serratus (Altenbach and Hermanson, 1987). Nevertheless, in the cited paper concatenated depression of the humerus and scapula in the first half of the downstroke was confirmed by high-speed filming of the three species of bats on a leash and with steel pins implanted into these bones and sticking out from under the skin. However, the method of analysis was not described in detail, and the data obtained with the help of this technology needs to be checked using fluoroscopy.

Vaughan argued that in addition to this lock employed at the upper-most wing position, the wing action in flight is also affected by the lock between supraglenoid tubercle of scapula and the groove between greater and lesser tuberosities of humerus. According to Vaughan (1959) this lock comes into action to confine protraction at extreme forward position of the humerus.

Further, Vaughan's ideas of locking mechanisms in the shoulder joint of bats were developed by Kovtun (1978). In addition to the so-called “upper locking,” described by Vaughan, he proposed for horseshoe bats the mechanism of “lower locking” between the lesser tuberosity of humerus and the coracoid process of scapula at the final phase of downstroke. In the wing beat cycle of bats Kovtun identified two modes: the free movement of the humerus (in subhorizontal wing positions) and the joint movement of the humerus and scapula (when the wing approaches its upper-most and lower-most positions), which he regarded as a very important destinctive feature of bat flight.

According to the study of syndesmological preparations, available mobility in the shoulder joint of Hipposideros hardly differs from that of Rousettus (Figs. 4-6). This allows us to conclude that the so-called “locking mechanisms” hypertrophied in Hipposideros actually do not constrain the mobility at all.

The peak observed in the caudo-ventral part of the extreme elbow trajectory in Hipposideros (Figs. 4,9) could be explained by the action of Kovtun's “lower locking.” However, no specialized articular surfaces on the humeral lesser tuberosity or scapular coracoid process, which are necessary for Kovtun's “lower locking,” were found neither by Schlosser-Sturm and Schliemann (1995), nor by ourselves in the same species which were studied by Kovtun. Therefore, the observed restriction of depression (adduction) in the caudal part of elbow trajectory in Hipposideros can be attributed not to any bony struts, but to winding of dorsal part of joint capsule around the humeral head. This part of the capsule is strengthened by the terminal tendon of infraspinatus muscle, which is probably more tight in Hipposideros than in Rousettus.

The range of humerus movements in Rousettus during flight is much narrower than the potential one on syndesmological preparations and employs only the most posterior part of the latter (Fig. 9). In flight, the elbow remains always well behind the shoulder joint and the humerus keeps an enormous reserve of unused protraction. There is good reason to believe that other bats in flight also use only the caudal-most part of the potential humerus relative to scapula mobility range. On one hand, experiments with syndesmological preparations (Figs. 4-6) show the close similarity of the potential mobility of the humerus in shoulder joint in Hipposideros having hypertrophied “locking” structures with that of Rousettus completely devoid of them. Conversely, the video records (Fig. 10) show the basic similarity of flight movements in these animals. Finally, the elbow movements of flying Rousettus according to X-ray data fall inside the ultimate trajectory of elbow of syndesmologic Hipposideros (Fig. 9). All of this allows us to conclude that the flight movements of the humerus in the shoulder joint which we found in Rousettus would generally hold for various groups of bats. It remains unclear in what situations bats protract humerus anterior to glenoid, in which position the so-called locks could act, since this requires simultaneous protraction of hind limb connected to the fore limb by wing membrane.

Figure 9.

Trajectories of elbow, right side view. Pink points—actual trajectory of R. aegyptiacus in flight according to X-ray video (arrows—direction of movement), blue and red lines—potential trajectories of R. aegyptiacus and H. armiger, respectively, according to experiments with syndesmological preparations (Fig. 4). Trajectories are scaled to equal humerus length and superimposed by the center of shoulder joint, the planes of scapulae oriented horizontally. The minor changes in the sagittal projection of the humerus due to latero-medial turning of scapula in X-ray sequence are not taken into account.

Figure 10.

Successive stages of wing beat cycle in free flight. (A) R. leschenaulti, (B) H. armiger. High speed (300 fps) video records of the same flights are provided in Movies S3, S4 in Supporting Information.

The shape and size of humeral tuberosities are associated with the lever ratios of attached muscles. For example, the configuration of the greater tuberosity determines the balance between supination caused by infraspinatus muscle and protraction caused by supraspinatus muscle. In case the greater tuberosity protrudes further from the humeral head, the maintenance of the humerus circumduction range requires development of corresponding notch on scapula. Otherwise the tuberosity would collide with the dorsal edge of glenoid earlier than in bats with less developed tuberosity, and this collision would result in significantly different trajectory of the humerus circumduction, which difference is not observed (Figs. 4-6). Thus, in our opinion the facets near glenoid are not blocking the shoulder, but on the contrary serve as notches for humeral tuberosities, which help to avoid restriction of mobility. In addition, if any lock operates in the most loaded part of the stroke, as is presumed in the “locking mechanisms concept,” then its surface area must be comparable to the area of glenoid itself. The fact that the area of glenoid is much larger implies that the main force is transmitted through it rather than through additional facets. Even if in some bats scapula and humerus are adducted downwards together (Altenbach and Hermanson, 1987), we believe that it is caused by muscular coordination only as soon as the humerus in flight is never protracted enough for locking of supraglenoid tubercle of scapula with the groove between greater and lesser tuberosities of humerus, as well as for locking of greater tuberosity of humerus with the additional articular facet of scapula.

Finally, it is noteworthy to compare the mobility of the shoulder girdle of bats uncovered above with that of terrestrial quadrupedal therians (Kuznetsov, 1985, 1995, 1999; Fischer, 1999; Fischer and Blickhan, 2006). Contrary to bats, dorsocaudal angle of their scapula does not shift considerably relative to the thorax while walking and running. This angle acts like a fixed pivot about which the scapula turns fore and aft in the parasagittal plane (similar to the femoral head in the hind limb). As to the clavicle, if present, it simply follows the swings of the acromion. No axial rotation of the clavicle was ever reported for terrestrial quadrupedal therians. In general, the shoulder girdle mobility in bats is more restricted in amplitude but is more complicated, and the clavicle plays a more important role, since it has become a vertical strut for the scapula displaced to the horizontal plane from the parasagittal one. The more detailed discussion of the evolution of the bat shoulder girdle will be soon published in our monograph entitled “Flight of Mammals: From Terrestrial Limbs to Wings” (Panyutina et al., 2012).


All the discovered degrees of freedom of scapula and clavicle are making a very minor contribution to the wing movement as compared to the input of humerus movements relative to scapula in the shoulder joint.

The main movement of the clavicle is its rotation around the straight proximal part (the stem). The scapula movements are confined to the frontal plane. As a whole, the shoulder girdle moves relative to thorax as a crank mechanism: clavicle plays the role of crank, and scapula—the role of connecting rod. These movements are out of phase with the downstroke and upstroke movements of humerus, although the cycle duration is the same.

In the shoulder joint only the most posterior part of the potential range of humerus mobility relative to scapula is used in flight. During downstroke humerus moves in a more protracted position than during upstroke.

Action of “locking mechanisms” and combined dorso-ventral movements of humerus and scapula are not supported by experimental data. In contrast to humerus, scapula moves mainly in the frontal plane.


Invaluable contribution to the technical support of the experimental part of work was made by S.M. Forsunov, N.M. Mylov, O.G. Ilchenko, M.А. Bragin, A.A.Tupikin. The authors are grateful to E.L. Yakhontov, F.Y. Dzerzhinsky and E.G. Potapova for their help in the process of interpretation of the results and writing the manuscript. T. Strickler and A. Borisenko helped by supplying with some rare publications. Video filming was facilitated by the Russian-Vietnamese Tropical Centre IEEP RAS.