Homologization of the Flight Musculature of Zygoptera (Insecta: Odonata) and Neoptera (Insecta)

Among the winged insects (Pterygota) the Dragonflies and Damselflies (Odonata) are unique for several reasons. Behaviourally they are aerial predators that hunt and catch their prey in flight, only. Morphologically the flight apparatus of Odonata is significantly different from what is found in the remaining Pterygota. However, to understand the phylogenetic relationships of winged insects and the origin and evolution of insect flight in general, it is essential to know how the elements of the odonatan flight apparatus relate to those of the other Pterygota. Here we present a comprehensive, comparative morphological investigation of the thoracic flight musculature of damselflies (Zygoptera). Based on our new data we propose a homologization scheme for the thoracic musculature throughout Pterygota. The new homology hypotheses will allow for future comparative work and especially for phylogenetic analyses using characters of the thoracic musculature throughout all winged insects. This will contribute to understand the early evolution of pterygote insects and their basal phylogenetic relationship.


Introduction
Within the insects the Odonata arguably are the group with the most impressive flight skills (e.g. [1]). Each wing pair can be controlled independently and some species are even able to fly backwards [2]. Through these flight skills Odonata are the avian key predators among insects [1].
The unique flight abilities are also reflected in a unique morphology. The meso-and metathorax forms a functional unit, the ptero-or synthorax, which is tilted caudally by 45u. The pleurites are strongly enlarged in dorso-ventral direction, whereas, the tergites and sternites are unusually small if compared to other pterygotes [2][3][4].
The muscles responsible for the wing movement are connected via cap tendons and sclerites directly to the wings [5]. This exclusively direct mechanism of wing movement distinctly sets Odonata apart from all other winged insects; where the wing beat is done mainly through a system of indirect muscles, many of which are highly reduced or missing in the Odonata (e.g. [6]).
Several publications address the structures of the flight apparatus of Odonata [5,[7][8][9][10], the aerodynamics of odonatan flight [10][11][12], the mechanics [2] and function of the flight musculature and the mechanoreceptors of the wing [10] as well as the complexity of the wing venation [13]. All these publications deal mainly with representatives of Anisoptera. In total, the knowledge about the odonatan thorax morphology shows a distinct deficit for the Zygoptera, which we, therefore, focused our comparative investigation on.
Presently there seems to be widespread agreement on ground pattern hypotheses for the wing base sclerites and for the flight musculature in Neoptera [15][16][17][18]. Even homologies between Ephemeroptera and Neoptera are mainly resolved [17,19], while hypotheses on the homologies between Odonata and the remaining Pterygota are still under discussion [17,19,14,10].
The aim of our comprehensive comparative investigation of the flight musculature of the Zygoptera is to identify variabilities among the Zygoptera and to establish homology hypotheses for the thoracic musculature of Odonata and Neoptera.
Together with the description of the muscles found, we already present our homology hypothesis by using the muscle names as proposed for Neoptera by Friedrich & Beutel [18]. We are aware that this presents a mixture of description and interpretation. However, stricter separation of these aspects would not support a clear and easily understandable presentation of the results.
For the skeletal elements of the thorax the nomenclature by Asahina [7] is used. Where necessary, this is supplemented by Snodgrass [6] and Ninomiya and Yoshizawa [14].
The homologies as well as the presence or absence of each muscle are listed in Table 1. In the muscle descriptions Asahina's muscle numbers are given in square brackets after the name of each muscle. For mesothoracic muscles Asahina's numbers for the corresponding metathoracic muscles are added in parentheses. The muscles are listed due to their occurrence in the pterothorax, from anterior to posterior. An additional table comparing our results with data from several other publications is available as supporting information (Table S1).
Since the prothorax has no active role in flight, it is omitted in this study.

Musculature of the Pterothorax
In the following we describe 44 muscles, 19 muscles of the mesothorax and 23 muscles of the metathorax. Two previously undescribed muscles, M. mesopleura-scutalis proximalis (IItpm2) and M. metapleura-scutalis proximalis (IIItpm2), are described for P. nymphula, C. puella, I. elegans, E. cyathigerum and P. latipes. The presence of these two muscles in P. pennipes could not be confirmed.
Origin: Preepisternum 2. P. nymphula (Fig. 4), C. splendens (Fig. 10). Insertion: Inserted with a long tendon at the anterior edge of proximale costal plate two (pCP2). The point of insertion is not exactly the edge but rather the membrane, which is connected with pCP2.
P. nymphula (Fig. 3, 4), Characteristics: The muscle is short and thin and has a dorsal cap tendon. It is a direct tonic depressor muscle [10].
P. nymphula (Fig. 2). Characteristics: It is a short muscle distal from muscle IIdvm3. Its presence in P. pennipes could not be confirmed. The muscle is elongate in P. latipes compared to P. nymphula. In C. splendens it shows its maximal length.
Origin: Basal at the anterior part of the mesocoxa. In C. splendens more postero-lateral at the base of the mesocoxa. P. nymphula (Fig. 1).
Insertion: Inserted with a tendon at the lateral part of the semidetached scutal plate. In C. splendens at the upper edge of pCP2.
P. nymphula (Fig. 3 Origin: Basal at the anterior part of the mesocoxa, posteromedian of muscle IIdvm4. In C. splendens caudal of muscle IIdvm4. P. nymphula (Fig. 1). Insertion: With a tendon at the proximal edge of axillary plate two (AxP2), or at the membrane between the mesoscutum and AxP2.
P.nymphula (Fig. 3), C. splendens (Fig. 10). Characteristics: The muscles IIdvm4, IIdvm5 and IItpm9 are direct flight muscles. IIdvm4 and IIdvm5 are tonic lifters and IItpm9 is a phasic depressor [10]. IIdvm4 and IIdvm5 have cap tendons and are distinctly smaller than IItpm9. The insertion in C. splendens is located laterally and more ventral than in P. latipes. In all species studied the origin of muscles IIdvm4 and IIdvm5 is also the attachment point of the coxal musculature.
P.nymphula (Fig. 4, 5). Characteristics: This muscle is strongly developed in E. cyathigerum and in I. elegans. It is missing in C. splendens and L. viridis. Its presence in P. pennipes could not be confirmed. This muscle was not described for Odonata so far. It is thin and elongate and runs almost parallel to muscle IIpcm1. It assumes a similar function as IIpcm1 and/or is reinforcing it.
IItpm9 -M. mesepimero-axillaris tertius [29/30 (51/52)]. Origin: With a short cap tendon at the posterior pleural process. P. nymphula (Fig. 6B). Insertion: In longitudinal axis at the ventral part of AxP2, precisely at the internal, caudal side of AxP2 next to the base of the anal vein. P.nymphula (Fig. 6B, C). Characteristics: IItpm4 and IItpm9 are located close together at AxP2, between the epifulcrum and the dorsal sclerite. Muscle IItpm4 is stronger and located more ventral; both have a cranial cap tendon. In C. splendens these muscles are distinctly separated from each other.
Characteristics: This muscle has a cap tendon and runs similar to IItpm9, but in comparison it is distinctly smaller. It is a direct depressor muscle [10]. Origin: In the middle at the pleural segmental border between meso-and metathorax.
P. nymphula (Fig. 1, 5). Characteristics: A short and thin muscle, with a dorsal, small cap tendon. The muscle is attached through resilin at the dorsal end [10].
Origin: Similar to muscle IIpcm4, at the lateral side of the mesocoxa.
P. nymphula (Fig. 1, 4). Characteristics: A short muscles with cap tendons at both ends. These cap tendons are each attached to the cuticle through long tendons.
Characteristics: This muscle corresponds to muscle IIdvm1 of the mesothorax, but is distinctly stronger. The presence in P. pennipes could not be confirmed. In C. spelndens the muscle is thin and elongate, whereas, in E. elegans it is quite small. It is missing in Epiophlebia [7].
P. nymphula (Fig. 7B,C), C. splendens (Fig. 11). Characteristics: IIIdvm4 and IIIdvm5 share the same point of origin at the coxa 3. Each muscle inserts via a long tendon. They have dorsal cap tendons and serve as direct flight muscles.
Characteristics: This muscle inserts at the lateral wall of the apodem where also muscle IIIdvm3 is attached. It is an indirect tonic flight muscle [10]. The muscle is stronger than its relative in the mesothorax.
IIItpm2 -M. metapleura-praealaris (new muscle). Origin: Dorsal region of the pleural bar between episternum 3 and epimeron 3, dorsal of muscle IIItpm9. P. nymphula (Fig. 5). Insertion: Median semi-detached scutal plate. Characteristics: This muscle is strongly developed in E. cyathigerum and in I. elegans but it is missing in C. splendens and L. viridis. Its presence in P. pennipes could not be confirmed. It is thin and elongate, runs almost parallel to IIpcm1 and was not described in Odonata so far. It assumes a similar function as muscle IIpcm1 and/or is reinforcing it.
Characteristics: In C. splendens this muscle is distinctly flatened caudally.
P. nymphula (Fig. 3), C. splendens (Fig. 10). Characteristics: This muscle has a dorsal cap tendon that inserts directly at the antecosta. The muscle is broader and shorter than in Anisoptera [10]. The cap tendon in I. elegans inserts ventral of muscle IIIscm6. In L. viridis the cap tendon is weakly developed.

Discussion
Asahina [7] listed 51 thorax muscles for Odonata in general and 42 muscles for adult Zygoptera. Of these muscles 19 belong to the mesothorax with Zygoptera lacking muscles 35 and 37. The remaining 23 muscles belong to the metathorax where Zygoptera do not have muscle 47, 57 and 60. In our study the 42 muscle of Asahina [7] could be confirmed. Additionally, four muscles (IIpcm2, IItpm2, IIIpcm2, IIItpm2) were found that were previously not known for the Odonata (cf. [5,7,10,23]). IIpcm2 and IIIpcm2 are present in all species studied, with the exception of L. viridis and C. splendens. In P. pennipes the condition is uncertain because of insufficient data.
The short and slender pleuro tergal muscles IItpm2 and IIItpm2 are run from the dorsal part of the pleural bar to the median semidetached scutal plate (Fig. 7A). They have positions and directions similar to IItpm4 and IIItpm4. Therfore, we assume a similar or reinforcing function (cf. [10]).
A couple of observed origin and insertion points differ from Asahina's [7] descriptions. For example, IIdlm1 inserts at the anterior edge of the postnotum 2, not at the lateral side of the scutum 3 [7]. The muscles IIdlm1, IIIdlm1 and IIIdlm2 have been identified as indirect flight muscles [24]. They originate at the tergal apophysis and were previously homologized with dorsal longitudinal muscles of the neopteran pterothorax [10]. In the ground pattern of the Neoptera the longitudinal muscles run between the phragmata [3]. The point of insertion of muscle IIdlm1 at the caudal edge of the postnotum, i.e. at the caudal end of the second thorax segment, is equivalent to the position of the phragma in Neoptera, which supports the homologization proposed.
In C. splendens IIdvm4 and IIdvm5 originate at the distal base of the mesocoxa (cf. [7]). In the other seven species investigated, these muscles originate rather cranial at the anterior part of the mesocoxa. Since Mnais strigata, which was studied by Asahina [7] and C. spledens both belong to Calopterygidae, the translocation of the point of origin may well be an apomorphy of this group.
The points of origin of the corresponding metathorax muscles IIIdvm4 and IIIdvm5 differ from previous descriptions [7] in all species investigated. They are located caudal not distal of the base of the metacoxa.
Further more, IIdvm4 has been described as attaching to the inner caudal angle of the costal plate 2. In the Zygoptera investigated, IIdvm4, like its metathoracic homolog IIIdvm4, is attached to the lateral side of the semi-detached scutal plate. The muscles do not attach at the wing articulation, rather at a tergal sclerite. Therefore, they have to be characterized as indirect not as direct flight muscles [10]. This also applies to the strong indirect lifter IIdvm3 (and IIIdvm3), which is a main flight muscle and is also attached to the tergum.
Consequently, the flight musculature of the Zygoptera consists of direct and historically indirect flight muscles. However, as far as the functions of the dorso-ventrally arranged flight muscles are concerned, all are now acting as direct muscles.

Homology of the Musculature of the Pterothorax in Zygoptera and Neoptera
Already in the descriptive part of this work we used the muscle nomenclature suggested by Friedrich and Beutel [18] for a generalized neopteran thorax. In the following the homologization of the flight musculature of Zygoptera with that of Neoptera is explained further (cf. Table 1, S1).
Dorsolongitudinal musculature (dlm). The tergal apophyses are intersegmental invaginations and therefore not homologous to the primary diaphragms of Neoptera [24], but presumably to the pseudo phragmata of other insects [25]. The zygopteran muscles IIdlm1, IIIdlm1 and IIIdlm2 originate at the tergal apophysis and their homology with the dorsolongitudinal musculature of Neoptera appears to be unequivocal [17].
Dorsoventral musculature (dvm). The points of origin and insertion of the zygopteran dorsoventral muscles are usually shifted to some degree in comparision to Neoptera. The reasons for this are not so much functional modifications, but drastic changes in shape and size of the notum of Odonata in coparision to that of other Pterygota. Nevertheless, the functions of these muscles as elevators of the wings are preserved. Their positions in the thorax together with the relationships to other muscles allow for a wellsupported homologization. The muscles IIdvm1 and IIdvm3, IIdvm4, IIdvm5 could be identified in the odonatan thorax.
Ventral musculature (vlm). The ventral muscle system in the Zygoptera appears to be highly simplified. We could identify one unequivocal ventral longitudinal muscle only: Ivlm7 is identical in its origin and insertion to its neopteran relative [18]. It seems not to be present in the Anisoptera but was also found in the Ephemeroptera [17].
Tergopleural musculature (tpm). The muscles IItpm6 and IItpm2 originate dorsally at the pleural bar. Muscle IItpm6 inserts below the proximal region of the axillary plate. In Neoptera IItpm6 inserts on the 3. axillary. The proximal area of the odonatan axillary plate has been homologized with the 3. axillary of Neoptera [14], which supports our identification of this muscle.
IItpm2 inserts on the median semi-detached scutal plate. Therefore, a homology with either the neopteran IItpm2 or IItpm4 seems to be possible.
An identification of this muscle as IItpm4 could be excluded, because in Neoptera IItpm4 inserts on the 1. axillary [18], which in Odonata corresponds to the anterior-proximal area of the axillary plate [14]. Since IItpm2 inserts on the subtegula or on the prealare sclerite in Neoptera, which correspond to the odonatan scutal plate, our homologization appears to be most probable.
The points of origin of IItpm9 and IIItpm9 at the pleural processes of their segments as well as the points of insertion on the axillary plates (homologous region see above) correspond well to the situation in the Neoptera and also in the Ephemeroptera [17].
Due to the virtually identical points of origin and insertion in the Neoptera [18] as well as in the Odonata the homologization of the metathoracic muscles IIItpm4 and IIItpm6 appears to be unequivocal.
Pleuro-coxal musculature (pcm). The zygopteran muscles IIpcm1 and IIIpcm1 originate at the preepisternum of the corresponding segments at the anterior edge of the pCP. Due to the ventro-dorsal expansion of the pleura in Odonata, this sclerite is directed nearly ventrally. Therefore, the orientations of the muscles in the thorax differ from there relatives in the Neoptera.
However, the points of origin and insertion together with the relation to other muscles support the homologization.
The zygopteran muscles IIpcm4, IIIpcm4, IIpcm6 and IIIpcm6 show the same points of origin and insertion as their neopteran counterparts. Together with functional considerations this supports the suggested homologization. Nevertheless, there is some variation in the points of insertion of IIpcm6 and IIIpcm6. In Zygoptera they insert on the trochanter of the corresponding segments, very close the insertion of IIpcm4 or IIIpcm4, respectively. In Anisoptera and in Epiophlebia these insertions are shifted to some degree [7].
In summary, our comparative investigation of the flight musculature of the Odonata shows that homologization with the flight musculature of Neoptera in most cases is realtively straightforward. Due to the significant modifications of the skeleton of the odonatan pterothorax many points of origin shifted in variing degrees. However, the general positions and orientations of the muscles are still persistent. It also became clear that the flight musculature of Zygoptera and of Odonata in general is composed of direct as well as indirect muscles as it is the case in the Neoptera. Those muscles that historically are indirect flight muscles work as direct flight muscles in the Odonata due to the modifications in their skelettal system, especially in the notal sclerites. With a well-supported homologization of the flight muscles between the Zygoptera (and consequently Odonata) and the Neoptera, this character system now can also be used to expand datasets for the analysis of phylogenetic relationships of all pterygote insects.
All regulations concerning the protection of free-living species were followed.
All necessary permits were obtained for collecting Odonata at the Billingshä user Schlucht, Göttingen, Germany (permission granted by ''Untere Naturschutzbehörde'' file reference AZ.67.2.5 Wei). For collecting Damselflies in France, no specific permits are required. The locations where the damselflies were collected are not privately owned or protected in any way. No endangered or especially protected species were collected.
Specimens were studied, prepared and drawn with the help of a stereomicroscope (Zeiss Stemi SV11) with a camera lucida.
Synchrotron radiation micro computed tomography (SRmCT) was applied in order to generate data for three-dimensional reconstruction of the structures of interest. Prior to scanning, the samples were critical point dried (Balzer CPD030). The SRmCT data were generated at the Swiss Light Source (SLS) in Villigen Three-dimensional reconstructions (processing and visualization) of the data were prepared with Amira H 5.2. (Visage Imaging, Richmond, Australia). All images were subsequently processed with Photoshop CS3 (Adobe System Inc., San José, USA).