Cruising specialists and accelerators – Are different types of fish locomotion driven by differently structured myosepta?

Zoology 106 (2003): 203–222 © by Urban & Fischer Verlag http://www.urbanfischer.de/journals/zoology

Cruising specialists and accelerators – Are different types of fish locomotion driven by differently structured myosepta? Sven Gemballa* and Kerstin Treiber Evolution and Bio-Geosphere Dynamics Program (EBID), Systematic Zoology, University of Tübingen, Germany Received May 15, 2003 · Revised version received July 29, 2003 · Accepted August 16, 2003

Summary Locomotor specialists, such as accelerators and cruisers, have clearly differing body designs. For physical reasons these designs are mutually exclusive, i.e. cruisers necessarily have poor accelerating capabilities and vice versa. For the first time, we examine whether differences in the anatomy of the musculo-tendinous system of the trunk are present in addition to the differences in external body design. We investigated the myoseptal series of two closely related locomotor specialists, the cruiser Scomber scombrus and the accelerator Channa obscura, by microdissections combined with polarized light microscopy and histology. Our comparison includes 3D-morphology of myosepta, spatial arrangement and length of myoseptal tendons, their relation to red and white muscles, rostrocaudal changes in all these aspects and the musculo-tendinous system of the caudal fin. Regarding all these features, Channa has retained the plesiomorphic condition of its actinopterygian ancestor. In contrast, the derived morphology of Scomber is characterized by (i) lateral (LT) and myorhabdoid tendons (MT) that are lengthened to up to 20% of body length (compared to a maximum of 8.2% in Channa), (ii) posterior myoseptal cones that are subsequently linked by horizontal projections of merged LTs and MTs, (iii) an increased area of red muscle fibers that insert to LTs of myosepta, (iv) the reduction of epineural (ENTs) and epipleural tendons (EPTs) that connect backbone and skin, (v) specific caudal tendons that are identified to be serial homologues of LTs and MTs of more anterior myosepta, (vi) and a partial reduction of intrinsic caudal muscles. These results suggest the following functional adaptations in the cruiser Scomber. Red muscle forces may be transmitted through LTs and posterior cones to the prominent tendons of the caudal fin. The length of LTs and the intersegmental connections along the posterior cones may facilitate posterior force transmission and may be correlated with the long propulsive wavelength generally observed in cruising carangiform swimmers. Epineural and epipleural tendons are interpreted to minimize lateral backbone displacement during high body curvatures. This is consistent with the lack of these tendons in Scomber, because high body curvatures are not displayed in stiffer-bodied carangiform swimmers. It remains to be tested whether the specializations revealed in this initial study for Scomber represent general specializations of carangiform swimmers. Taking into account the geometry of myoseptal tendons and the horizontal septum we evaluate how local bending according to beam-theory can be generated by white or red muscle activity in Channa and Scomber. In both species, the musculo-tendinous anatomy of the caudal fin explains the functional asymmetry of the caudal fin that was experimentally revealed in previous studies. Key words: Channa, Scomber, myoseptal tendons, swimming, force transmission, red muscle, white muscle, carangiform, bending, caudal musculature

Introduction Several attempts have been made to embrace the range of locomotor diversity in fishes by definition of locomotion types. Based on initial definitions given by

Breder (1926), the main criteria for differentiation of locomotion types (e.g. anguilliform, subcarangiform, carangiform, thunniform) were obtained from swimming parameters (e.g. Lindsey, 1978; Videler, 1993). An alternative classification of locomotor diversity in

*Corresponding author: Sven Gemballa, Evolution and Bio-Geosphere Dynamics Program (EBID), Department of Systematic Zoology, University of Tübingen, Auf der Morgenstelle 28, D-72076 Tübingen, Germany; phone: ++49-7071-2976947; fax: ++49-7071-295150; e-mail: [email protected] 0944-2006/03/106/03-203 $ 15.00/0

S. Gemballa and K. Treiber

fishes was derived from the integration of hydrodynamics, external locomotor morphology (‘body design’), modelling, and ecology (Webb, 1984a, b; Webb, 1988; Weihs, 1989). This approach revealed three main types of body designs (maneuverer, accelerator, and cruiser) that are each optimized for certain behaviors and habitats, and are mutually exclusive due to physical and hydrodynamic reasons. That is, each type is specialized for one swimming function but performs poorly in the other two functions. These specialists are at extreme positions of a continuum of locomotor morphologies. Generalists represent an intermediate form with respect to both external morphology as well as their ability to perform cruising, accelerating, or maneuvering. This study examines the internal locomotor morphology of two of the three defined locomotor specializations, a cruising and an accelerating specialist. Despite the large body of knowledge on swimming capabilities of fishes we have only limited knowledge of internal mechanisms of power swimming movements using complex three-dimensionally arranged myomeres. Although several issues, such as muscle physiology, body mechanics, and modelling, have been addressed (e.g. Altringham and Ellerby, 1999; Ellerby and Altringham, 2001; Katz et al., 1999; Jayne and Lauder, 1995; Long, 1998; Long et al., 1994; Long and Nipper, 1996; Long et al., 2002; Mc Henry et al., 1995; Rome et al., 1993; Shadwick et al., 1999), we are still far from a thorough understanding of myomere mechanics and force transmission in swimming fishes. Some light has been shed on internal myomere mechanics by recent anatomical studies (Gemballa and Vogel, 2002; Gemballa et al., 2003a) that revealed an evolutionarily conserved set of myoseptal tendons. These tendons might play a key role in myomere mechanics (e.g. force transmission or modulation of body stiffness). However, most of this research focussed on locomotor generalists. So far, it has never been investigated whether locomotor specialists differ not only in external morphology but also show marked differences in their internal morphology, and probably also in the function of myomeres. We compare the myomeric morphology of a cruiser, the mackerel Scomber scombrus, to that of an accelerator, the snakehead Channa obscura. For scombrid fishes, comparative data on myomere morphology are available from Westneat et al. (1993). Mackerels (Fig. 1B) match the optimized body design of a cruising specialist, i.e. a high aspect ratio lunate tail, a narrow caudal peduncle, a relatively stiff and streamlined anterior body, and large anterior muscle mass (Fierstine and Walters, 1968; Webb, 1984; Westneat and Wainwright, 2001). In terms of tail beat amplitudes, wavelengths and body bending, their swimming mode is clearly carangiform (Wardle and Videler, 1993; Donley and 204

Dickson, 2000). Snakeheads (Fig. 1A) are specialized accelerators characterized by a large body depth compared to body length, a large body and fin area, especially in the posterior region (compare to Scomber), a large muscle mass, and high values of extreme body curvatures (Webb, 1984; Gemballa and Bartsch, 2002 for Channa). Judging from physiological data, accelerators and cruisers should differ considerably in their pathways of force transmission. In cruisers, the bulk of musculature is located anteriorly, where almost no undulations occur, but thrust is applied to the water posteriorly at the lunate tail. It is likely that anteriorly generated forces are transferred posteriorly through the narrow caudal peduncle to the lunate tail. In contrast, accelerators can make use of local muscles in all parts of the body to generate undulations all along the body (Wardle et al., 1995). The mechanism of posterior force transmission in cruisers remains unclear. One hypothesis suggests that posterior muscle fibers serve as force transmitters in cruisers. It has been frequently observed that posterior muscle fibers (in contrast to anterior ones) are active while being lengthened on the convex side of the body (for review see Wardle et al., 1995; Gillis, 1998; Altringham and Ellerby, 1999). This eccentric muscle activity is most pronounced in cruisers, and occurs to a lesser extent in non-cruisers (e.g. generalists or accelerators). Whether muscle fibers with eccentric activity perform negative work and thus serve as force transmitters is controversial (Ellerby and Altringham, 2001; Gillis, 1998; Katz and Shadwick, 1998; Shadwick et al., 1998, 1999). A second hypothesis is that myosepta play a relevant role in force transmission (Nursall, 1956; Wainwright, 1983). This hypothesis had long been neglected but was recently supported by the finding that longitudinally arranged tendons are present in all parts of myosepta (Gemballa et al., 2003a; Gemballa and Röder [in press], Gemballa [in press]., Gemballa and Hagen, unpubl.). Given that cruisers, in contrast to non-cruisers, transmit anteriorly generated forces to the tail, we predict that posterior myoseptal tendons of cruisers are designed for posterior force transmission. Though the mechanism remains unknown to date, one way to fulfil this function might be longer posterior myoseptal tendons in relation to the anterior ones. In contrast to cruisers, we expect that myoseptal morphology in accelerators is more uniform along the body, because it allows for local bending at all axial positions. Given this framework, we have to consider the whole series of myosepta of a cruiser and an accelerator rather than individual myosepta. Our analysis includes assessment of the myoseptal 3D-morphology, collagen fiber architecture, relationship of myoseptal tendons to red Zoology 106 (2003) 3

Myosepta of cruisers and accelerators

Fig. 1. Body shape and basic myoseptal anatomy. Body shape and skeleton of Channa obscura (A) and Scomber scombrus (B). Framed areas indicate regions that were investigated histologically. (C) Schematic drawing of a myoseptum and underlying axial skeleton of Channa obscura. Lateral view, anterior to left. The horizontal bar at the bottom illustrates how the rostrocaudal extension is measured. Dashed line = medial line of attachment of myoseptum (MLA) to axial structures, solid white line = lateral line of attachment of myoseptum to skin; black horizontal bar = rostrocaudal extension of MLA; grey horizontal bar = length of dorsal anterior cone. Length of dorsal posterior cone is zero. (D), (E) Medial lines of attachment (black) mapped onto axial skeleton (grey; vertebrae 1–34) in Channa obscura (D) [each 5th myoseptum between 3 and 28] and Scomber scombrus (E) [vertebrae 2–28; MS8, 14, 21]. Notice attachment to anterior margin of vertebral centrum and horizontal course across three neural arches in (D), and three neural spines in (E). For abbreviations see Fig. 2. Zoology 106 (2003) 3

205

S. Gemballa and K. Treiber

and white muscles and axial skeleton, length measurements of myoseptal tendons, and assessment of changes of myoseptal structure along the body. Data on the horizontal septum of both snakeheads and mackerels are obtained from previous studies (Gemballa and Britz, 1998; Gemballa et al., 2003b). Differences identified in the myoseptal system along the trunk of the cruiser Scomber and the accelerator Channa and between the two species may clarify patterns of myoseptal functions in the two types of swimming. Such an analysis of functional patterns has to consider the evolutionary history of the structures that contribute to a certain function (Lauder, 1990, 1991, 1995). Generally, derived structures (evolutionary novelties) are correlated with functional adaptations. In the functional context of this study, it is important to know whether features of the myoseptal system in Channa and Scomber represent derived (apomorphic) or plesiomorphic features of one of the two genera. Both genera are closely related perciform genera (Nelson, 1994; Helfman et al., 1997) and fortunately we have good knowledge about the perciform groundpattern from previous studies (Gemballa and Britz, 1998; Gemballa et al., 2003a). This groundpattern closely resembles that of higher clades (e.g. gnathostomes and actinopterygians) and was retained throughout the teleost radiation. Hence, differences in the myoseptal structure of any of the two investigated perciforms with the perciform groundpattern must have evolved within this clade and are likely to be related to the evolution of their locomotory specializations.

Material and methods Microdissections and polarized light microscopy of myoseptal series

We carefully skinned formalin-fixed specimens and afterwards cleared and stained them according to a modified procedure of Dingerkuhs and Uhler (1977). Details of the procedure including modifications and the following microdissections are described in Gemballa and Britz (1998) and Gemballa and Hagen (unpubl.) and are only briefly addressed here. After completion of the clearing and staining procedure pure ethanol was used as storage medium and as medium for the microdissections. After careful removal of endomysial tissue, myosepta were excised by cutting close to their insertion line along the horizontal septum and midsagittal plane using fine irisspring scissors (Vannas Mini Tübingen). A few myosepta were retained in their positions at various regions of the trunk. Their 3D-arrangement was documented by drawings prepared with a camera lucida system. Myosepta were given ascending numbers from anterior to posterior. In order to compare myoseptal features 206

along the trunk of each species and the myosepta of both species, we recorded several characteristics of their 3D-morphology: (1) the number of vertebral segments spanned by certain parts of a myoseptum, (2) the number of segments spanned by the medial attachment of a myoseptum to axial structures (Fig. 1C; dotted line), and (3) the number of segments spanned by the anterior and posterior cones. From these three measurements we obtained the rostrocaudal extension of a myoseptum (Fig. 1C). To facilitate interspecific comparisons, the values of rostrocaudal extension are expressed in percent of total length (% TL). In each of the specimens we studied the myoseptal series along the whole trunk up to the caudal fin. Our analysis included the modifications of posteriormost myosepta that directly insert on caudal fin rays (i.e. tendinous system of caudal fin musculature). Exact measurements of the rostrocaudal extension of myosepta are difficult, because the tips of the myoseptal cones are not quite fixed in their position after clearing and staining. We divided a vertebra into four quarters (0.0, 0.25, 0.50 and 0.75 of vertebral length) and assigned the tip of a cone to one of these quarters. Thus, the accuracy dropped to ±0.25 of vertebral length (corresponding to 0.35% TL in Channa and 0.56% TL in Scomber). In other words, accuracy of measurements of rostrocaudal extensions lies within 1.12% TL (± 0.56% TL). Excised myosepta were spread out on microscopic slides. Their collagen fiber architecture was visualized under a Zeiss Stemi 2000-C stereomicroscope equipped with a polarizer and analyzer (Zeiss Polarizer S and analyzer A53). Under polarized light, collagen fibers within myosepta appear as white strands, whereas the remaining parts of myosepta remain black. Micrographs were taken using either a MC-100 camera system (b/w films; 400 ASA) or a digital camera (Fujix DC HC-300Z; 1000 × 1500 pixel) adapted to the stereomicroscope. All myosepta and specimens are stored in pure ethanol at the zoology department in Tübingen (personal collection of SG). 3D-reconstructions and illustrations

We used computer based 3D-reconstructions (see Sánchez et al., 2001) of axial structures (horizontal septum, vertebral column and median septum) of the midbody to prepare 3D-greyscale artwork of the midbody. We chose oblique dorsal and anterior views for the computer model so that all myoseptal details could be viewed adequately. Since a complete 3D-reconstruction of a myoseptum was impossible to achieve by the computer software we obtained reference points from the microdissections (e.g. position of tips of cones, length of cones, insertion of myoseptum to axial structures) and added these manually into the 3D graph of the Zoology 106 (2003) 3

Myosepta of cruisers and accelerators

trunk. Initial line drawings of the 3D-model including the spatial arrangement of myosepta served as a basis to produce a 3D-greyscale artwork of the midbody region. Data on collagen fiber pathways within a myoseptum were obtained from polarized light microscopy, and were finally added to the greyscale artwork. In contrast to the midbody morphology, the 3D-morphology of the posterior body was documented by camera lucida drawings from lateral views of dissected specimens. Histological investigations

We used two of our specimens (Channa obscura 160 mm TL; Scomber scombrus 214 mm TL) for histological investigations. From each of the two specimens we obtained sagittal sections of 20 µm thickness at axial positions between 32%–45% TL and 67.5%–77.5% TL. The sections were stained according to the Azan-Domagk or the Azan-Heidenhain procedure. Selected sections were digitized using a Zeiss Standard microscope with a Fujix digital camera (HC-300Z; 1000 × 1500 pixel). Specimens examined

Microdissections were carried out on five specimens of Channa obscura (Günther, 1861) of varying lengths (114 mm, 139 mm, 143 mm, 148 mm, 157 mm TL) and three specimens of Scomber scombrus (Linné, 1758) (74 mm, 275 mm, 385 mm TL). Additionally, two specimens (see above) were used for histology. All specimens and histological series are stored at the department of Zoology (personal collection of SG) except for the 74 mm TL specimen of Scomber (American Museum of Natural History, AMNH 38043).

Results Spatial arrangement of myosepta: midbody and posterior body

A myoseptum is spanned between its two lines of attachment. Laterally, it attaches to the skin, medially, it attaches to axial structures, such as the vertebral column, median septum and peritoneum. A single myoseptum does not form a plane sheet between its two lines of attachment, but has a complex 3D-morphology due to the presence of myoseptal cones which are posterior and anterior prolongations into the musculature. Insertion of myosepta to axial structures

In Channa as well as in Scomber, the medial attachment line of a myoseptum has the form of a ‘W ’ turned by 90 ° (Fig. 1D, E). However, some marked differences between the two species are present in the epaxial as well as in the hypaxial parts. In Channa, an individual epaxial myoseptum runs dorsally along the anZoology 106 (2003) 3

terior margin of a vertebral centrum N. It turns caudally at the base of the neural arches (anterior body and midbody; Fig. 1D) or at the base of the neural spines (posterior body, Fig. 1D) to proceed almost horizontally across three vertebral segments (N, N + 1, N + 2). Along the third segment (N + 2) it follows the course of the neural spine to its distal end. Here, it takes a sharp turn towards a craniodorsal direction to end at the dorsal midline. Epaxial myosepta of Scomber also attach to the anterior margin of a vertebral centrum (N). In contrast to Channa, the attachment line proceeds along the neural spine of this vertebral segment (N). Its orientation alters towards a horizontal course only after approximately two-thirds of the neural spine have been covered (Fig. 1E). Three neural spines (N, N + 1, N + 2) are traversed almost horizontally until the attachment line turns craniodorsally to end at the dorsal midline. Hypaxially, myosepta insert at ventral ribs in the anterior and midbody of Channa and Scomber (Fig. 1D, E). At the distal tip of the ribs, the attachment line turns cranioventrally to end at the ventral midline. The hypaxial attachment line turns towards the posterior body. In Scomber, a dorsoventral symmetry of epaxial and hypaxial attachment line is gradually achieved (Fig. 1E). Here, the hypaxial attachment line traverses the hemal spine of three vertebral segments (N, N + 1, N + 2). In Channa, the position of the dorsal backward flexure is mirrored by the ventral backward flexure. However, the horizontal course of the epaxial attachment line across parts of three successive vertebral segments is not achieved in the hypaxial part, where the attachment line remains within one segment (along the ventral ribs). A condition of marked asymmetry between epaxial and hypaxial parts is present in the anterior body of Channa, where ventral ribs have shifted dorsally and hypaxial parts of myosepta are in a more vertical position. 3D-morphology of myosepta

Myosepta are drawn out into cones that project into the musculature. According to their position these cones are termed dorsal posterior cone (DPC), ventral posterior cone (VPC), dorsal anterior cone (DAC), and ventral anterior cone (VAC; Fig. 2). They are connected by myoseptal sheets. Following Gemballa et al. (2003a), we term these sheets epaxial sloping part (ESP; extending between DAC and DPC), hypaxial sloping part (HSP; extending between VAC and VPC), epaxial flanking part (EFP; extending between DPC and dorsal midline), and hypaxial flanking part (HFP; extending between VPC and ventral midline; Fig. 2). From the course of the medial attachment lines (Fig. 1D, E) it is evident that the medial portions of ESP parts are oriented almost horizontally and span three 207

S. Gemballa and K. Treiber

Fig. 2. 3D-morphology and collagen fiber architecture of midbody myosepta. Left side of body, anterolateral view. (A) Epaxial midbody region (49–63% of TL) of Channa obscura. Slightly dorsal view. Shape and collagen fiber architecture of three myosepta shown. Posterior one (MS23) is complete, middle one (MS22) almost completely excised. Dorsal fin rays cut. (B) Hypaxial midbody region (49–63% of TL) of Channa obscura. Slightly ventral view. Shape and collagen fiber architecture of MS23 is completely shown. Medial part of VAC of MS21 additionally shown. Anal fin rays cut. (C) Midbody region (47–64% TL) of Scomber scombrus. Shape and collagen fiber architecture of MS15 illustrated. The dorsoventral distribution of red muscle fibers along the lateral part of a myoseptum is indicated by the short horizontal lines that originate at the MS and point posteriorly. Myoseptal structures: DAC = dorsal anterior cone, DPC = dorsal posterior cone, EFP = epaxial flanking part, ESP = epaxial sloping part, HFP = hypaxial flanking part, HSP = hypaxial sloping part, VAC = ventral anterior cone, VPC = ventral posterior cone. Collagen fiber architecture of myoseptum: ENT = epineural tendon, LT = lateral tendon, MLF = mediolateral fibers, MT = myorhabdoid tendon. Axial structures: afm = anal fin musculature, dfm = dorsal fin musculature, hes = hemal spine, hs = horizontal septum, ms = median septum, na = neural arch, ns = neural spine, p = peritoneum, vc = vertebral centrum, vri = ventral rib.

208

Zoology 106 (2003) 3

Myosepta of cruisers and accelerators

Fig. 3. Channa obscura 148 mm TL. Polarized light micrographs of myosepta. Collagen fiber tracts visible as white strands. Myosepta were spread out after excision. Dashed line represents medial attachment line to axial structures. Myosepta were cut along cones (dotted lines) to allow for spreading out the four myoseptal parts separately. (A) Midbody myoseptum. MS17 at axial position of 46% TL. (B) Posterior myoseptum. MS34 at 70% TL. Notice dorsoventral symmetry of collagen fiber architecture. For abbreviations see Fig. 2. Scale bar represents 2% TL. Zoology 106 (2003) 3

209

S. Gemballa and K. Treiber

vertebral segments. Owing to this, ESPs of successive myosepta get very close to each other and even merge, forming horizontal layers of multiple sheets of ESPs. We term these horizontal layers ESP-multi-layers. They are difficult to separate by microdissection and are discernible in sagittal histological sections (Fig. 6 A, B). In Channa they extend along the neural arches (Fig. 1D), whereas in Scomber they exist at the distal level of the neural spines (Fig. 1E). Epaxial and hypaxial parts are almost symmetrical in terms of their 3D-shape in the midbody and posterior

body. This dorsoventral symmetry also applies to the medial attachment lines (see above) that show a similar course, and to the length and position of cones. However, symmetry is less pronounced in the anterior body region, where hypaxial cones are shorter than epaxial ones (and anteriormost even absent in Channa). This principal arrangement of myosepta is maintained along the body. However, some aspects of myoseptal morphology (e.g. length of myoseptal cones) alter gradually in a rostrocaudal gradient along the trunk (see paragraph on rostrocaudal changes below).

Fig. 4. Scomber scombrus 275 mm TL. Polarized light micrographs of myosepta. Collagen fiber tracts visible as white strands. Myosepta were spread out after excision. Dashed line represents medial attachment line to axial structures. Myosepta were cut along cones (dotted lines) to allow for spreading out the four myoseptal parts separately. (A) Midbody myoseptum. MS11 at axial position of 46% TL. Dorsoventral symmetry of collagen fiber architecture almost developed. (B) Posterior myoseptum. MS22 at 70% TL. Dorsoventral symmetry of collagen fiber architecture well developed. Notice increase in length of myoseptal parts between (A) and (B) For abbreviations see Fig. 2. Scale bar represents 2% TL.

210

Zoology 106 (2003) 3

Myosepta of cruisers and accelerators

Collagen fiber architecture of myosepta: midbody and posterior body

Myoseptal architecture and myoseptal tendons in Channa obscura In Channa each epaxial myoseptum contains a set of three tendons. These tendons contribute to the robustness of a myoseptum, whereas the remaining parts contain only few and relatively weak collagen fibers. Two of these tendons, the lateral and myorhabdoid tendon, are oriented longitudinally and connect myoseptal cones. The myorhabdoid tendon extends between the anterior tip of the EFP and the DPC. It is present in the lateral part of an EFP (Figs. 2A, 3A, B). A lateral tendon connects DAC and DPC. This tendon lies in the lateral part of an ESP (Figs. 2A, 3A, B). In its middle part it is connected to the skin. Both myorhabdoid and lateral tendon are intramuscular tendons that do not insert to skeletal or axial structures. In contrast, the third tendon of the epaxial part, the epineural tendon, firmly inserts to the neural arch of a vertebra (Figs. 2A, 3A, B). Along its caudolateral pathway across an ESP towards the skin its collagen fibers diverge markedly (Fig. 3A). Epineural tendons are part of the ESP-multi-layers and thus reinforce these horizontal layers in a caudolateral direction between vertebral axis and skin. Apart from the three tendons described there are only few collagen fibers that contribute to the robustness of an epaxial myoseptum. The most pronounced fibers extend between the median septum and the skin of the EFP and

are termed mediolateral fibers (Fig. 3A, B). The principal arrangement of the epaxial myoseptal tendons is maintained along the whole body (Fig. 3A, B). The three epaxial tendons are mirrored hypaxially by three hypaxial tendons (Fig. 3B). In the flanking parts (EFP and HFP) corresponding myorhabdoid tendons are present, in the sloping parts (ESP and HSP) corresponding lateral tendons are present. The epineural tendon of the ESP corresponds to the epipleural tendon of the HSP. However, this condition of dorsoventral symmetry is not true for the anterior body region which lacks distinct myoseptal tendons hypaxially (Fig. 3A). The HSP and HFP sheets of these myosepta consist of evenly distributed collagen fibers extending in two directions almost mediolateral to each other and crossing at acute angles. When tracing hypaxial myosepta from the anterior to the posterior body, myorhabdoid, lateral and epipleural tendons develop gradually. Fiber density appears to be slightly enhanced in the region where epipleural tendons are present in posterior myosepta (Fig. 3A). Myoseptal architecture and myoseptal tendons in Scomber scombrus In Scomber each epaxial myoseptum contains a set of two longitudinally arranged tendons (Fig. 4A, B; see also Fig. 2C). Clearly, these tendons represent the homologs of the lateral and myorhabdoid tendon in Channa. While lateral tendons that extend between anterior and posterior cones are broad and not very distinct, myorhabdoid tendons are generally more distinct

Fig. 5. Association of red muscle fibers and myosepta in Scomber scombrus. (A) Transverse section at 66% TL, left body. Grey area represents red muscle. Vertical line represents position of sagittal section depicted in (B). (B) Sagittal section of epaxial part, anterior to left. Position of horizontal septum indicated by dashed line. White muscle (anterior part of figure) consists of relatively thick fibers. Red muscle is visible in dorsalmost and posterior part. It consists of more slender and densely packed fibers. The posterior part of the section is situated more laterally, because the trunk tapers. In this part only red muscle is visible. The anterior part shows red muscle fibers in its dorsal part according to (A). (C) Detail of area indicated by box in (B). Note the two lateral tendons that are in association with white and red muscle fibers. Zoology 106 (2003) 3

211

S. Gemballa and K. Treiber

Histologically, red muscle fibers are easily discriminated from white fibers by their slender shape. As could be expected from locomotory behavior, we found very few red muscle fibers in Channa (too few for quantification but clearly below 1% of a section). All of them are situated laterally along the horizontal septum, wedged between white fibers. In contrast, Scomber has a considerable portion of red muscle (about 6.2% of total body mass, between 8% and 24% of mass in crosssections between 50% TL and 80% TL; Graham et al., 1983). It is not restricted to the region around the horizontal septum, but extends dorsally and ventrally to the level of the posterior cones (Figs. 2C, 5). Thus, in Scomber most of the red muscle fibers are not connected to the horizontal septum but insert into lateral parts of myosepta in which lateral tendons are present.

constant. In Channa, both DAC and DPC length slightly increase, whereas ESP length remains constant. Horizontal projections of myoseptal tendons are present at the DPCs. These projections are formed by the posterior ends of the merging myorhabdoid and lateral tendons. Thus, we included these projections into our measurements of rostrocaudal extension. In Channa, horizontal projections are not developed in the anterior body (Fig. 6D), are weakly pronounced in the midbody, and well pronounced in the posterior body (Fig. 6E). Here, they span about half the distance to the next myoseptum. In contrast, horizontal projections in Scomber are well pronounced along the whole body length and span almost three quarters of the distance between adjacent posterior cones (Fig. 6F, G). In addition to the increase of length in lateral tendons, we recorded an increase of length towards the posterior body in myorhabdoid tendons of the epaxial flanking parts. As in the case of the lateral tendons, this increase is marked in Scomber and moderate in Channa. In the latter, anterior EFPs span 5.5% TL (MS10 at an axial position of 36% TL), 5.9% TL in the midtrunk (MS21 at 51% TL), and 6.3% TL in the posterior body (MS31 at 66% TL). In Scomber, EFPs contain the longest tendons of all. Their maximum length in the posterior body is 19.1% TL (MS21 at 66% TL). Even the more anterior EFPs (MS8 at 36% TL spans 15.1% TL) and midbody EFPs (MS14 at 51% TL spans 17.1% TL) are remarkably long as compared to Channa.

Rostrocaudal changes within the series of myosepta

Musculoskeletal system of the posterior body

Two myoseptal features change markedly in a rostrocaudal gradient along the trunk. First, the length of myoseptal tendons increases towards the posterior body. Secondly, myosepta of the posterior body are linked by horizontal projections of myoseptal tendons. In Scomber, there is a marked increase in length of lateral tendons from midbody to posterior body as compared to Channa (Fig. 6C). These tendons connect anterior and posterior cones, thus spanning the whole myoseptal part from the anterior tip of the DAC to the posterior tip of the DPC. In Scomber, the body region spanned by the DAC, ESP, and DPC of a single myoseptum increases from 11.6% TL (MS12 at axial position of 47% TL) to 16% TL (MS25 at 75% TL). This relative increase by approximately 38% is not observed in Channa. In this species, a midbody myoseptum (MS16 at 46% TL) spans a region of 7.2% TL. The rostrocaudal extension increases to 8.2% TL in the posterior body (MS35 at 75% TL; relative increase of approximately 14%). Most of the increase in length in Scomber is caused by the almost threefold increase of DAC length (Fig. 6C: lower faded areas of bars). In contrast, the length of ESP and DPC remains almost

In Channa, the myoseptal morphology of the midbody is quite similar to that of the posterior myosepta with respect to the medial lines of attachment and the 3D-shape of myosepta (compare Fig. 7A, B to Figs. 1A, D, 2A, B). The posteriormost complete myoseptum attaches to the fourth from last vertebra (including the terminal pseudurostyle; see Day, 1914; Monod, 1968; Gosline, 1968 for skeletal anatomy of Channa and terminology). The myotomes posterior to this form the superficial flexor muscle (Fig. 7B). Its connective tissue consists of longitudinally directed collagen fibers that attach to the caudal fin rays. The dorsalmost and ventralmost fin rays are connected to derivatives of the flanking parts of myosepta, and the central fin rays are connected to derivatives of the ESPs and HSPs. This superficial layer of caudal fin musculature is separated from a deeper layer, the superficial flexor muscle (Fig. 7A; dashed line). The tendinous parts of this muscle are not attached to the underlying skeleton. They run within the muscle and insert to the caudal fin rays. The muscle itself covers the whole area of the caudal fin skeleton including the pseudurostyle (terminal vertebra). The whole portion of caudal musculature is sub-

and narrow in their middle part and fan out at both ends. In contrast to Channa, Scomber lacks distinct epineural or epipleural tendons. Instead, a number of mediolateral fibers is present in the posterior part of each ESP and HSP (Fig. 4A, B; see also Fig. 2C). A dorsoventral symmetry in the architecture of EFP and HFP is present not only in the posterior body, but also in the midbody. Although less conspicuous, the same is true for ESP and HSP (Fig. 4A, B; see also Fig. 2C). Association of red muscles and myoseptal tendons

212

Zoology 106 (2003) 3

Myosepta of cruisers and accelerators

divided into an epaxial and hypaxial half by the horizontal septum and the longitudinal hypochordal muscle. This muscle originates at the pseudurostyle and a hypural bone. It is oriented caudodorsally and its tendons insert to four caudal fin rays (Fig. 7B, dashed line). In Scomber, derivatives of myosepta pass through the narrow caudal peduncle. The lateral and myorhabdoid tendons of these myosepta form pronounced and elongated tendons. The medial parts of these myosepta are weak and it was impossible to reliably record their attachment sites along the axial skeleton. In our speci-

mens we found MS27 to be the posteriormost myoseptum in which myorhabdoid and lateral tendons merge to form a DPC and a horizontal projection. The latter inserts to fin rays in the central part of the caudal fin (Fig. 8A). Polarized light micrographs of the region between the anterior cone and the distal attachment clearly reveal that this tendon is a serial homolog of a lateral tendon of preceding myosepta. A weak medial part containing few mediolateral fibers is discernible and the whole collagen fiber architecture resembles that of posterior myosepta (compare Fig. 8C [upper

Fig. 6. Comparison of myoseptal series of Channa obscura and Scomber scombrus by sagittal histological sections and length measurements of myoseptal tendons. (A), (B) Horizontal course of ESPs close to the midsagittal plane. ESPs come in from an anteroventral direction and merge with a horizontal layer, the ESP-multi-layer (arrows). An ESP-multi-layer is formed by ESPs of successive myosepta. Left body side, anterior to left. (A) Channa obscura at axial position of 70% TL. (B) Scomber scombrus at axial position of 40% TL. Detailed view at right bottom. (C) Length measurements of lateral tendons of myosepta at different axial positions along the body of two specimens of Channa obscura (148 mm TL, 157 mm TL; orange bars) and Scomber scombrus (275 mm TL, 385 mm TL; blue bars). The fully coloured central area of each bar represents the area spanned by the medial line of attachment of a myoseptum. The lower faded area represents the area spanned by the anterior cone; the upper faded area represents the area spanned by the dorsal posterior cone or its horizontal projections into the trunk musculature. Values of % TL given at each bar. (E)–(G) Horizontal projections (HP) into the trunk musculature at the DPCs of different axial positions. (D), (E) Channa obscura. Midbody ((E); 40% TL) is devoid of any HPs, whereas an HP covers half the distance between adjacent cones in the posterior body ((F); 69% TL; arrows). (F), (G) Scomber scombrus. In both midbody ((F); 40% TL) and posterior body ((G); 69% TL) HPs (arrows) cover almost the full distance between adjacent DPCs. Zoology 106 (2003) 3

213

S. Gemballa and K. Treiber

Fig. 7. Channa obscura. Musculoskeletal system of the posterior body. Lateral view, anterior to left. (A) Medial attachment lines of myosepta (black lines) mapped on caudal skeleton (grey). Anteriormost vertebra is 33. Area covered by the Musculi flexor profundus dorsalis and ventralis (MPD, MPV) and their insertion to caudal fin rays indicated by dashed line; attachment area of Musculus hypochordalis longitudinalis (MHL) visible between MPD and MPF. (B) 3D-shape of MS33 and MS37 and flanking parts of MS33 to MS39. The area posterior to MS39 is occupied by the Musculus flexor superficialis (MFS). Insertion of MFS and MHL (dashed line) to caudal fin rays shown.

myoseptum] to ESP in Fig. 4B). MS27 is the first myoseptum that inserts to caudal fin rays, and is the longest of the caudal myosepta. Compared to preceding myosepta (e.g. MS25 spans 16.0% TL), its length has increased to up to 19.1% TL. Myosepta posterior to MS27 also end at caudal fin rays. Thus, the length of myoseptal tendons decreases in MS28 and MS29 (Fig. 8C). Posterior to MS 27, derivatives of MS28 and MS29 form two conspicuous caudal tendons. Both myosepta do not form posterior cones but have their EFP and ESP parts separately attached to caudal fin rays. Their flanking parts insert to the dorsalmost and ventralmost fin rays (Fig. 8A, B; only EFP of MS28 shown). The lateral tendon formed by the ESP of MS28 covers that of MS27 (Fig. 8A). The former has been frequently termed the medial caudal tendon in scombrids (Westneat et al., 1993; Westneat and Wainwright, 2001). The medial caudal tendon underlies the lateral tendon of MS29 (Fig. 8B), for which the term great lateral tendon (GLT) was used in scombrids (Westneat et al., 1993; Westneat and Wainwright, 2001). The posterior end of a single great lateral tendon diverges markedly, giving rise to two asymmetrical arms. However, the epaxial and hypaxial great lateral tendons together give the caudal fin a dorsoventral symmetry. In all of these myoseptal parts, longitudinally oriented collagen fibers are predominant (Fig. 8C). In addition to the described derivatives of myotomes, some intrinsic caudal muscles are present in Scomber (Fig. 8D). The longitudinal hypochordalis muscle underlies, and is interwoven with, the LT29 (GLT). Dis214

tally, it inserts to one of the upper caudal fin rays, and proximally it originates from the dorsal crest of the parhypurapophysis (Fig. 8D). At its posteroventral margin this bone gives rise to the hypaxial portion of the profound flexor muscle. The tendons within this muscle are oriented posteroventrally and insert to some of the lower caudal fin rays (Fig. 8D). The tendons running in the anteroventral part of the muscle appear to be very robust, while weakening towards the posteroventral part. Both longitudinal hypochordalis muscle and profound flexor muscle seem to be weakly developed in comparison to the homologous muscles of Channa. Another conspicuous tendon is oriented longitudinally and inserts to the parhypurapophysis. This deep tendon spans the last three vertebrae (Fig. 8D). Our investigations of the horizontal septum of the posterior body confirm the data given in the initial study of Westneat et al. (1993). The horizontal septum consists of an array of epicentral tendons (synonym: anterior oblique tendon; oriented posterolaterally) and posterior oblique tendons (POTs; oriented anterolaterally; see Fig. 3K of Gemballa et al., 2003b). Both are anchored at the vertebral column. It is important to note that the posteriormost epicentral tendon inserts to the seventh from last vertebra (at 75% TL), and the posteriormost POT inserts to the fifth from last vertebra (at 79% TL). Posterior to this, a distinct horizontal septum is missing, but tendinous fibers diverge from the mid horizontal plane and merge with the lateral tendons of the posteriormost myosepta LT27–29. Superficially, the muscular arrangement described so far is covered by a subdermal sheath of connective tissue. Zoology 106 (2003) 3

Myosepta of cruisers and accelerators

Fig. 8. Scomber scombrus. Musculoskeletal system of the posterior body from vertebrae 29 to caudal fin. Lateral and myorhabdoid tendons of MS27–29 pass through the narrow caudal peduncle to insert to caudal fin rays. Lateral view, anterior to left. (A) Epaxial and hypaxial lateral tendons of MS27, and epaxial and hypaxial lateral and myorhabdoid tendons of MS28. (B) Epaxial and hypaxial lateral tendons of MS29 added to B. (C) Polarized light micrographs of epaxial sloping parts (ESPs) of MS27 (upper), MS28 (middle), and MS29 (lower micrograph). ESPs consist almost exclusively of lateral tendons. Notice similarity of MS27 with MS22 (Fig. 4B). Scale bar represents 2% TL. (D) Intrinsic caudal muscles inserting to the parhypurapophysis (phu). Extension of muscles is represented by grey areas, tendons within muscles are indicated by darker grey areas. MHL = Musculus hypochordalis longitudinalis, MFP = Musculus flexor profundus; DT = deep tendon.

Zoology 106 (2003) 3

215

S. Gemballa and K. Treiber

Discussion Comparison of the myoseptal series of Channa, Scomber, and gnathostome fishes

Taking into account previous studies on the myoseptal system (Gemballa and Britz, 1998; Gemballa et al., 2003a) we are able to decide whether the features of the myoseptal system described here for the perciform genera Channa and Scomber represent derived (apomorphic) or plesiomorphic features when compared to the perciform ancestor. In this section we mainly identify the novelties that evolved in the myoseptal series and caudal fin of the two genera. It is crucial to identify these specialized structures before analyzing whether

Fig. 9. Simplified phylogenetic diagram showing the pattern of 11 features of the myoseptal system and caudal fin that are addressed in this study. Black symbols indicate derived features, white symbols plesiomorphic features. Note that Channa has completely retained all of the perciform groundpattern whereas Scomber is derived with respect to features 6B–11B. The features 1A–11A of the perciform ancestor represent plesiomorphic features that were retained either from the gnathostome or actinopterygian or teleost ancestor. References [1] Gemballa et al. 2003a, [2] Gemballa and Röder (in press), [3] Gemballa and Britz 1998, [4] Gemballa et al. 2003b, [5] Lauder 1989, [6] Gemballa, in press, [7] Gibb et al. 1999, [8] Westneat and Wainwright 2001.

216

they may be functionally related to the locomotory specializations (accelerating or cruising). Fig. 9 includes a list of 9 features of the myoseptal system that were present in the assumed perciform ancestor. These features closely match those of the myoseptal system described for Channa. In contrast, Scomber is derived in four respects (features 6–9): (1) It lacks epineural/epipleural tendons; (2) the longitudinal tendons (lateral and myorhabdoid tendons) are elongated; (3) the ESP-multi-layers are shifted dorsally to the level of the neural spines, and (4) horizontal projections are more pronounced. The most remarkable feature is the elongation of the lateral and myorhabdoid tendons. Unfortunately, there are only few comparative data from basal actinopterygins (Gemballa and Röder, in press). For example, lateral tendons of Polypterus span 6.8–8.4% TL, those of Acipenser 5.2–7.8% TL, and those of Lepisosteus 7.3–9.1% TL. These data are in good accordance with those of Channa (7.2–8.2% TL; Fig. 6C) but differ markedly from those of Scomber (11.6–19.1% TL). Although comparative measurements of myoseptal tendon lengths within the Scombridae do not exist, there is some indirect evidence that these tendons are generally elongated in this family. Fierstine and Walters (1968) reported that the length of anterior cones changes remarkably in the pacific mackerel (Scomber japonicus). In addition, their counts of concentric rings of myosepta that appear in transverse section indicate that cones in more derived scombrids are even longer than in mackerels. In accordance with this, Knower et al. (1999) reported that myomeres in Thunnus albacares and Katsuwonus pelamis span 13 and 17 segments, respectively. A similar impression is derived from the illustrations of Westneat et al. (1993) and Westneat and Wainwright (2001) for various scombrid species. Given these findings, we consider elongated lateral tendons to be a unique feature of scombrids. In terms of the caudal fin musculature (Fig. 9, features 10–11), Channa has retained all plesiomorphic features, whereas in Scomber the longitudinal hypochordalis muscle appears to be weakly developed compared to perciforms (Gibb et al., 1999; Westneat and Wainwright, 2001) and the profound flexor muscle is developed hypaxially only. The deep tendon inserting to the parhypurapophysis (Fig. 8D) of Scomber is documented for the first time in this study. So far there is no indication of a homologous structure in other teleosts. Functions of myoseptal tendons: integration of anatomical and physiological data

An evaluation of the functional role of myosepta has to consider myoseptal tendons as well as their association with myomeric muscles. In Channa, myomeric muscles Zoology 106 (2003) 3

Myosepta of cruisers and accelerators

almost exclusively consist of white muscles, whereas in Scomber a considerable amount of red muscle is present laterally (Fig. 5A, see also Graham, 1983). These findings are in accordance with functional differences of the two muscle types as demonstrated by studies of locomotory habits and muscle physiology. According to these studies, red fibers are used for steady swimming, whereas white fibers are used additionally or exclusively during sprint swimming or escape respones (e.g. Boddeke et al., 1959; Rayner and Keenan, 1967; Rome et al., 1988; Jayne and Lauder, 1993, 1994; Gillis, 1998; Hale et al., 2002). For mackerels, this functional difference has been proven by electromyography (Wardle and Videler, 1993; Shadwick et al., 1998). No such data exist for snakeheads (Channa). Since these fishes almost completely lack red fibers, it is reasonable to assume they use white muscles for most of their activities. Given this, our results on the myoseptal system of Channa have to be discussed in the context of white muscle function. Because Scomber has a relatively high amount of red muscle for cruising and also a considerable amount of white muscle for bursts, the myoseptal anatomy of this species has to be discussed in the context of both red and white muscle function. Myoseptal design and white muscle activity

The main question to be addressed in this section is how white muscle fibers may interact with myoseptal tendons and whether there are functional differences in Channa and Scomber. Two possible ways of white muscle and tendon interaction were identified in a previous study (Gemballa and Vogel, 2002). Put briefly, epineural and epipleural tendons do not play a relevant role in transmission of white muscle forces, because they are at obtuse angles to the deeper white muscle fibers. Instead, these tendons are important for the generation of intramuscular pressure. In contrast, the lateral and myorhabdoid tendons are likely to transmit muscular forces, because the muscle fibers are at acute angles to these tendons (Gemballa and Vogel, 2002). Judging from our data on the length of lateral tendons it seems possible that these tendons transmit forces over some distance. The lateral tendons in Channa span between 7.2% and 8.2% TL which is close to the data obtained for basal actinopterygians (see above, Gemballa and Röder [in press]). In Scomber, tendons are longer (11.6% and 19.1% TL) and forces might be transmitted over a longer distance. If deeper white muscle fibers pull at the anterior end of a lateral tendon and the force is transmitted via the tendon to its posterior end, bending would have to be observed at a posterior position. Indeed, posterior bending was observed and the presence of caudal force transmitters was postulated from Zoology 106 (2003) 3

sonomicrometry and kinematic analysis of fast-starts in trout (Covell et al., 1991). In contrast to posterior bending, local bending might occur if all white muscle fibers along a lateral tendon from an anterior to a posterior cone contract simultaneously (e.g. contraction of the helical muscle fiber arrangement [HMFA] described by Gemballa and Vogel, 2002). Such local bending was demonstrated by sonomicrometry and kinematic analysis in milkfish during sprints (Katz et al., 1999). At present, a more detailed evaluation of the function of lateral and myorhabdoid tendons during white muscle activity is difficult, because too little is known about different patterns of muscle fiber recruitment within a myomere in different swimming types. The data available so far (Jayne and Lauder, 1995; Ellerby and Altringham, 2001) indicate that recruitment patterns are far more complex than previously thought. Given this, it remains unsolved whether transmission of white muscle forces in species with shorter tendons (e.g. Channa) is different from transmission of white muscle forces in species with longer tendons (e.g. Scomber). Comparative electromyography and sonomicrometry data that consider different muscle portions within one myomere and in myomeres of different axial positions are required to address this question. Force transmission in fishes is not only discussed with respect to single or adjacent myomeres, but also regarding transmission from anterior to posterior myomeres. This concept of posterior force transmission postulates that forces generated in anterior muscles are transferred posteriorly. In particular, white muscle fibers of posterior myotomes were hypothesized to transmit forces, since they were found to be active while being lengthened. While some studies are in favour of this, others present conflicting evidence for white muscles (Jayne and Lauder, 1993; Johnston et al., 1995; Ellerby and Altringham, 2001; Ellerby et al., 2001; d’Aout et al., 2001). In contrast to this ongoing debate, we propose an involvement of lateral and myorhabdoid tendons in posterior force transmission, instead of or in addition to white muscle fibers. Generally, their mechanical properties make tendons a more effective transmitter of eccentric activity than muscles. Our results suggest the following pathways for posterior force transmission: Forces are transferred along the lateral and myorhabdoid tendons to the posterior cones. At these cones the myoseptal tendons project into the musculature, and forces may be transferred to subsequent myosepta through these horizontal projections (Fig. 6E–G) and muscle fibers. If forces from rostral myotomes accumulate along these pathways, it is to be expected that these intermyoseptal connections are more pronounced in posterior myomeres than in anterior ones. Indeed, this is the case in both Channa and Scomber. In particular, 217

S. Gemballa and K. Treiber

posterior force transmission should be relevant for fishes with a high percentage of body musculature located anteriorly, and narrow caudal peduncles at the caudal fin that produce most of the thrust. This is generally the case for a cruiser. Consistently, the cruiser (Scomber) has longer tendons (Fig. 6C) and more pronounced horizontal projections (Fig. 6D–G) when compared to the accelerator (Channa). In terms of the function of epineural and epipleural tendons, our data are in accordance with the hypothesis that these tendons contribute to the generation of intramuscular pressure during bending (Gemballa and Vogel, 2002). While bending, radial swelling of contracting muscles on the concave side causes a lateral excursion of the backbone out of the midsagittal plane (Wainwright, 1983). Tendinous sheaths of myosepta that surround the swelling muscle resist the radial expansion and generate intramuscular pressure (Westneat et al., 1998). We propose that epineural and epipleural tendons represent important elements of this tendinous sheath, because they are part of the myoseptal sheath that wraps around contracting muscles of anterior myomere cones and forms mechanical linkages between backbone and skin. Thus, these tendons and the underlying white muscles form a muscular-hydrostat (see Kier and Smith, 1985 for biomechanical principles). Large body curvatures during fast-starts will lead to larger radial swellings of muscles and hence larger intramuscular pressure than moderate curvatures during cruising. Indeed, it has been shown in basal actinopterygians and one shark that enormous intramuscular pressure is produced during large body curvatures in fast-starts (Westneat et al., 1998: 60–90 kPA; Wainwright et al., 1978: 200 kPa) while low intramuscular pressure is developed during slight body curvatures in sustained swimming (Wainwright et al., 1978: 20–35 kPa). Assuming that epineural and epipleural tendons act as circular antagonists of swelling muscles, we expect that these tendons will be well developed in fast-starters such as Channa, which have to deal with greater bending curvatures. The tendons should be less pronounced in cruisers such as Scomber that show relatively small bending curvatures. Indeed, our data prove this correlation. Channa has well pronounced epineural and epipleural tendons (Fig. 3), while these tendons are formed by some evenly distributed collagen fibers (Fig. 4A) or almost lacking (Fig. 4B) in the cruiser Scomber. The lack of epineural and epipleural tendons in the cruiser Scomber is exceptional among gnathostomes. Generally, epineural and epipleural tendons are well pronounced in locomotor generalists (Gemballa et al., 2003a) and in the accelerator Channa (this study). Further studies will be necessary to test whether this lack 218

of tendons applies to cruisers in general. Interestingly, results on the carangid Trachurus mediterraneus (Gemballa and Hannich, unpublished data) revealed that these tendons are missing in the posterior body of this cruiser as well. Myoseptal design and red muscle activity

Virtually no red muscle fibers were present in Channa. Therefore, we confine the discussion to the analysis of pathways of red muscle forces in Scomber. The current hypothesis of red muscle force transmission was derived from a comparative analysis of seven scombrid species (Westneat et al., 1993). The major result of this study was that the horizontal septum is likely to be the main transmitter of muscle forces. Put briefly, bending is caused by red muscles pulling on the posterior oblique tendons (POTs) of the horizontal septum (in the following referred to as the POT-pathway of force transmission). Our observations of the horizontal septum in Scomber and the data from Gemballa et al. (2003b) match the anatomical data given by Westneat et al. (1993) for the horizontal septum of Scomber. The POT-pathway may be criticized for the following reason. The last POT inserts to the fifth from last vertebra in Scomber (Westneat et al., 1993, this study) and posterior to that the horizontal septum is missing. Accordingly, no force will be transmitted to the caudal fin or any of the caudal tendons or intrinsic caudal muscles (Fig. 8) by the proposed mechanism. This expectation regarding the POT-pathway is in contrast to direct force measurements in two species of tunas. These measurements indicate that forces are transmitted through the caudal tendons during steady as well as burst swimming (Knower et al., 1999). In Scomber, we identified tendons posterior to the horizontal septum that diverge from the mid horizontal plane and merge with the lateral tendons of the posteriormost myosepta LT27–29. This may be one mechanism to transmit red muscle forces to the caudal tendons. However, our data also suggest an additional mechanism that involves myoseptal tendons. So far, analysis of the transmission of red muscle forces has never taken into account myoseptal tendons that are associated with red muscles. Our data on the mackerel allow us to do so for the first time, and based on these data we are able to propose an alternative mechanism for force transmission, the lateral-tendon-pathway (Fig. 10). We demonstrated that red muscle fibers form a band of remarkable dorsoventral width along the body (Figs. 2C, 5; see also measurements of Shadwick et al. [1998] for S. japonicus) and are directly connected to parts of the lateral tendons of myosepta. Thus, muscular forces may be transmitted posteriorly by latZoology 106 (2003) 3

Myosepta of cruisers and accelerators

Fig. 10. Schematic representation of the proposed pathway of posterior force transmission of red muscle in the cruiser Scomber. Lateral view of left side. Posterior body with vertebrae 15–31 shown. Epaxial and hypaxial lateral tendons of MS17, MS21 and MS25 are completely shown, posterior ends of remaining lateral tendons are additionally shown. Anterior ends of lateral tendons are at anterior myoseptal cones. The posterior ends lie at the dorsal and ventral posterior cones of myosepta. Horizontal bars in epaxial part represent orientation of red muscle fibers. Arrows in hypaxial part indicate pathways of force transmission. The two posterior arrows indicate the medial and great lateral tendon of MS28 and MS29 (Fig. 8). If red muscle fibers pull on lateral tendons, forces will be transmitted along lateral tendons towards their dorsal and ventral posterior cones. Subsequent cones are mechanically linked by horizontal fanlike projections (see Fig. 6) and forces can be transmitted along the cones towards the caudal fin.

eral tendons and their dorsal posterior cones (DPCs). Since subsequent DPCs are mechanically linked by horizontal fanlike projections (Fig. 6 F,G) forces will be transferred further posteriorly towards the caudal tendons (Fig. 10). Both the POT-pathway and the lateral-tendon-pathway of force transmission are not exclusive. They may both be used and their importance may vary with swimming speed if, for example, only some part of the red muscles would be active at lower speeds. So far, differential activation of red muscles in swimming mackerels has not been examined. However, it has been shown in tunas that caudal tendons transmit forces during steady swimming (red muscle activity). Not surprisingly, the forces measured in caudal tendons increase 10-fold during burst activity (Shadwick et al., 2002). According to our results, this clearly indicates that caudal tendons transmit lower forces produced by fewer red muscle fibers during cruising and higher forces produced by white muscles during burst activity. The arrangement of tendons within the POT- and lateral-tendon-pathway will determine how muscular contraction is translated into bending. The shortest distance for which red muscle forces will be transmitted can be estimated for both pathways. From the geometry of the POTs in Scomber we infer that red muscle forces will travel a distance of 1.75 (midbody) to 3 (posterior body) vertebral lengths. This is equivalent to Zoology 106 (2003) 3

4–6.4% of TL. Bending will occur within this distance. If body curvatures are most pronounced in the middle of this distance they occur close to active red muscle (approximately 2–3% TL). The distance covered by the lateral tendon from its superficial contact with red muscles to the posterior cone is 2–2.5 vertebral lengths (see Fig. 2C; 5–5.5% TL). Thus, the lateral-tendonpathway can also be assumed to occur close to active red muscle (approximately 2.5–2.75% TL). This estimation is close to expectations resulting from the beam-theory (Katz et al., 1999). Indeed, red muscle strain and midline curvature have been shown to be in phase for some teleosts (including the chub mackerel S. japonicus) and one shark (Coughlin et al., 1996; Shadwick et al., 1998; Katz et al., 1999; Donley and Shadwick, 2003). Our hypothesis that red muscles may transmit forces through the long lateral tendons of myosepta (see Fig. 10) may be of broader importance with respect to the understanding of posterior force transmission in steady swimming. Posterior force transmission is important for cruising species because they generate most of the thrust at the caudal fin, but possess most of their musculature anterior to the narrow caudal peduncle. For example, in Scomber (as in other scombrids) the highest absolute area of red muscle is present at 50% TL (Ellerby et al., 2000). For some time posterior red muscles were believed to serve as force transmitters in 219

S. Gemballa and K. Treiber

carangiform swimmers, because they seemed to be active while being lengthened and were assumed to produce negative work (Altringham et al., 1993; Wardle and Videler, 1993; Wardle et al., 1995). According to evidence presented in other studies this functional role of posterior muscles seems to be overestimated. Net positive work is likely to be produced at all axial positions (e.g. Rome et al., 1993; Shadwick et al., 1998; Ellerby et al., 2000; Syme and Shadwick, 2002; d’Aout et al., 2001) indicating that tendons instead of muscles are crucial for force transmission. For the first time, we describe myoseptal tendons that are suitable for transferring red muscle forces. We also demonstrate that the serial homologs of these tendons are the conspicuous caudal tendons. Nevertheless, posterior muscles are likely to produce negative work during the initial phase of their activity (see summary of rostrocaudal shift of EMG muscle strain phase in Gillis [1998]). This might help to mechanically couple adjacent posterior cones of myosepta and thus to form intersegmental linkages – consisting of lateral tendons and active red muscle fibers – that can transmit forces. Given the characteristics of burst duration of anterior and posterior red muscles in carangiform swimmers (Gillis, 1998), this red muscle/tendon arrangement is simultaneously active along one side of the body for brief periods at a time. In addition to the muscle activity, the length of the tendons in this arrangement may contribute to the long propulsive wavelengths that are generally observed in more stiff-bodied carangiform swimmers. In contrast, anguilliform and subcarangiform swimmers (Gemballa and Röder [in press]) as well as the accelerator Channa possess shorter tendons and shorter propulsive wavelengths. Only recently, the importance of such intersegmental tendinous linkages has been demonstrated by computational modelling of fish swimming (Long et al., 2002). It is important to note that such linkages can transfer forces generated anteriorly by red muscle fibers to the caudal fin only if the red muscle fibers are connected to these tendons. Here, this relationship was demonstrated to exist in the mackerel, but we still lack data on other species. We also completely lack data on the relationship of internalized red muscles to myoseptal tendons. Such information might help to understand some of the unique characteristics that have been revealed for thunniform swimmers with internalized red muscles (Shadwick et al., 1999; Altringham and Shadwick, 2001; Katz et al., 2001; Katz, 2002). Functions of the caudal fin

All features of the caudal fin described for Channa were identified to be plesiomorphic for percomorphs (Fig. 9; features 10, 11). Some functional roles of these 220

muscles have been addressed experimentally in the percomorph Lepomis macrochirus (Lauder, 2000). The morphological similarity of the caudal fin in Lepomis and Channa suggests similarity in function. One main aspect is that the externally symmetrical caudal fin functions asymmetrically, with the dorsal lobe undergoing greater lateral excursions than the ventral lobe (see Lauder, 2000). Surprisingly, it has been shown that the caudal fin of the chub mackerel Scomber japonicus also functions asymmetrically (Gibb et al., 1999; Nauen and Lauder, 2002). In this case, not only the dorsal lobe of the caudal fin functions asymmetrically but also the caudal peduncle. Two mechanisms were hypothesized to be responsible for this functional asymmetry. First, activity of an intrinsic caudal muscle, the longitudinal hypochordalis muscle, may cause asymmetry, although this muscle seems to be poorly developed in Scomber (Fig. 8D). Secondly, asymmetry may also be caused by the myomeric musculature of the caudal peduncle (Gibb et al., 1999). Both hypotheses have not yet been tested experimentally in Scomber, and only the former was proven for one percomorph (Lauder, 2000). Based on our anatomical data it seems possible that both mechanisms contribute to functional asymmetry. The great lateral tendon (LT29; Fig. 8B,C) is clearly asymmetrical with respect to the mid horizontal plane. Greater lateral excursions of the tail are likely to be achieved when higher forces act on the epaxial LT29 (and probably the MT28) than on the hypaxial LT29 (and MT28). Asymmetrical forces may be the result of asymmetric activity of epaxial and hypaxial muscles, and this can be tested experimentally. Concluding remarks

This study revealed marked differences in the myoseptal system of an accelerator and a cruiser. While the myoseptal system of the accelerator resembled that of a locomotor generalist, the myoseptal system of the cruiser showed specializations, such as reduction of epineural and epipleural tendons and lengthening of myorhabdoid and lateral tendons. From these differences in structure and the different functional demands of accelerating and cruising, we inferred that epineural and epipleural tendons are important for generation of intramuscular pressure, whereas lengthened myoseptal tendons function as posterior force transmitters while cruising at long propulsive wavelengths. Such comparative studies on locomotor specialists may provide useful information on myoseptal functions. However, whether the conclusions presented in this initial study generally hold true for cruisers and accelerators should be tested in further comparative and experimental work. Zoology 106 (2003) 3

Myosepta of cruisers and accelerators

Acknowledgements We thank Wolfgang Maier (Tübingen) for advice and encouragement throughout the duration of this project. Jeanine M. Donley (San Diego) and two anonymous reviewers kindly helped to improve the manuscript. Technical support was provided by Thi Thi Fussnegger (histology) and Gabi Schmid (artwork on mackerel). Thanks to Barbara Brown (AMNH) for the loan of study material.

Literature Altringham, J.D., D.J. Ellerby. 1999. Fish swimming: patterns in muscle function. J. Exp. Biol. 202: 3397–3403. Altringham, J.D. and R.E. Shadwick. 2002. Swimming and muscle function. In: Fish Physiology 19: Tuna-Physiology, Ecology, and Evolution (B.A. Block and E.D. Stevens, eds.). Academic Press, San Diego. Pp. 314–344. Altringham, J.D., J.S Wardle and C.I. Smith. 1993. Myotomal muscle function at different locations in the body of a swimming fish. J. Exp. Biol. 182: 191–206. Boddeke, R., E.J. Slijper and A. van der Stelt. 1959. Histological characteristics of the body musculature of fishes in connection with their mode of life. Proc. K. Ned. Akad. Wet. Ser. C 62: 576–588. Breder, C.M. 1926. The locomotion of fishes. Zoologica (N.Y.) 4: 159–29. Coughlin, D.J., L. Valdes and L.C. Rome. 1996. Muscle length changes during swimming in scup: sonomicrometry verifies the anatomical high-speed cine technique. J. Exp. Biol. 199: 459–463. Covell, J.W., M. Smith, D.G. Harper and R.W. Blake. 1991. Skeletal muscle deformation in the lateral muscle of the intact rainbow trout Onchorynchus mykiss during fast start maneuvers. J. Exp. Biol. 156: 453–466. Day, A.L. 1914. The osseus system of Ophiocephalus striatus. Bloch. Phil. J. Sci., Sec. D. 9: 19–55. D’Aout, K.D., N.A. Curtin, T.L. Williams and P. Aerts. 2001. Mechanical properties of red and white swimming muscles as a function of the position along the body of the eel Anguilla anguilla. J. Exp. Biol. 204: 2221–2230. Donley, J.M. and K.A. Dickson. 2000. Swimming kinematics of juvenile kawakawa tuna (Euthynnus affinis) and chub mackerel (Scomber japonicus). J. Exp. Biol. 203: 3103–3116. Donley, J.M. and R.E. Shadwick. 2003. Steady swimming muscle dynamics in the leopard shark Triakis semifasciata. J. Exp. Biol. 206: 1117–1126. Dingerkus, G. and D.L. Uhler. 1977. Enzyme clearing of Alcian blue stained whole small vertebrates for demonstration of cartilage. Stain Technology 32: 229–231. Ellerby D.J. and J.D. Altringham. 2001. Spatial variation in fast muscle function of the Rainbow Trout Oncorhynchus mykiss during fast-starts and sprinting. J. Exp. Biol. 204: 2239–2250. Ellerby D.J., J.D. Altringham, T. Williams, T. and B.A. Block. 2000. Slow muscle function of Pacific bonito (Sarda chiliensis) during steady swimming. J. Exp. Biol. 203: 2001–2013. Fierstine, H.L. and V. Walters. 1968. Studies in locomotion and anatomy of scombroid fishes. Mem. South Cal. Acad. Sci. 6: 1–31. Zoology 106 (2003) 3

Gemballa, S. and P. Bartsch. 2002. Architecture of the integument in lower teleostomes: functional morphology and evolutionary implications. J. Morph. 253: 290–309. Gemballa, S. and R. Britz. 1998. The homology of intermuscular bones in acanthomorph fishes. Am. Mus. Nov. 3241: 1–25. Gemballa, S., L. Ebmeyer, K. Hagen, T. Hannich, K. Hoja, M. Rolf, K. Treiber, F. Vogel and G.W. Weitbrecht. 2003a. Evolutionary transformations of myoseptal tendons in gnathostomes. Proc. Roy. Soc. Lond. B 270: 1229–1235 Gemballa, S., K. Hagen, K. Röder, M. Rolf and K. Treiber. 2003b. Structure and evolution of the horizontal septum in vertebrates. J. Evol. Biol. 16: 966–975. Gemballa, S. and K. Röder (in press). From head to tail: the myoseptal system in basal actinopterygians. J. Morphol. Gemballa, S. (in press). Musculoskeletal system of the caudal fin in basal actinopterygians: heterocercy, diphycercy, homocercy. Zoomorphol. Gemballa, S. and F. Vogel. 2002. Spatial arrangement of white muscle fibers and myoseptal tendons in fishes. Comp. Biochem. Physiol. A 133: 1013–1037. Gibb, A.C., K.A. Dickson and G.V Lauder. 1999. Tail kinematics of the chub mackerel Scomber japonicus: testing the homocercal tail model of fish propulsion. J. Exp. Biol. 202: 2433–2447. Gillis, G.B. 1998. Neuromuscular control of anguilliform locomotion: patterns of red and white muscle activity during swimming in the American Eel Anguilla rostrata. J. Exp. Biol. 201: 3245–3256. Gosline, W.A. 1968. The suborders of perciform fishes. Proc. US Nat. Mus., Smithson. Inst. Washington D.C, 124: 1–78. Graham, J.B., F.J. Koehrn and K.A. Dickson. 1983. Distribution and relative proportions of red muscle in scombrid fishes: consequences of body size and relationships to locomotion and endothermy. Can. J. Zool. 61: 2087–2096. Hale, M.E., J.H. Jr. Long, M.J. Mc Henry, M.W. Westneat. 2002. Evolution of behavior and neural control of the fast-start escape response. Evolution 56: 993–1007. Helfman, G.S., B.B. Collette and D.E. 1997. The diversity of fishes. Blackwell Science: Malden, Massachusetts. Jayne, B.C. and G.V. Lauder. 1993. Red and white muscle activity and kinematics of the escape response of the bluegill sunfish during swimming. J. comp. Physiol. A 173: 495–508. Jayne, B.C. and G.V. Lauder. 1994. How swimming fish use slow and fast muscle fibers: implications for models of vertebrate muscle recruitment. J. Comp. Physiol. A 175: 123–131. Jayne, B.C. and G.V. Lauder. 1995. Are muscle fibres within fish myotomes activated synchronously? Patterns of recruitment within deep myomeric musculature during swimming in largemouth bass. J. Exp. Biol. 198: 805–815. Johnston, I.A., J.L. van Leeuwen, M.L.F. Davies and T. Beddow. 1995. How fish power predation fast-starts. J. Exp. Biol. 198: 1851–1861. Katz, S.L. 2002. Design of heterothermic muscle in fish. J. Exp. Biol. 205: 2251–2266. Katz, S.L., R.E. Shadwick and H.S. Rapoport. 1999. Muscle strain histories in swimming milkfish in steady and sprinting gaits. J. Exp. Biol. 202: 529–541. Katz, S.L. and R.E. Shadwick. 1998. Curvature of swimming fish midlines as an index of muscle strain suggests swimming muscle produces net positive work. J. Theor. Biol. 193: 243–256. Katz, S.L., D.A. Syme and R.E. Shadwick. 2001. Enhanced power in yellowfin tuna. Nature 410: 770–771. Kier, W.M. and K.K. Smith. 1985. Tongues, tentacles and trunks: the biomechanics of movement in muscular-hydrostats. Zool. J. Linn. Soc. 83: 307–324. Knower, T., R.E. Shadwick, S.L. Katz, J.B. Graham and C.S. Wardle. 1999. Red muscle activation patterns in yellowfin

221

S. Gemballa and K. Treiber

(Thunnus albacares) and skipjack (Katsuwonus pelamis) tunas during steady swimming. J. exp. Biol. 202: 2127–2138. Lauder, G.V. 1989. Caudal fin locomotion in ray-finned fishes: historical and functional analyses Amer. Zool. 29: 85–102. Lauder, G.V. 1990. Functional morphology and systematics: studying functional patterns in an historical context. Annu. Rev. Ecol. Syst. 21: 317–340. Lauder, G.V. 1991. Biomechanics and evolution: integrating physical and historical biology in the study of complex systems. In: Biomechanics in Evolution (J.M.V Rayner and R.J. Wootton, eds). Cambridge University Press, pp. 1–19. Lauder, G.V. 1995. On the inference of function from structure. In: Functional Morphology in Vertebrate Paleontology (J.J. Thomason, ed). Cambridge University Press, pp. 1–18. Lauder, G.V. 2000. Function of the caudal fin during locomotion in fishes: kinematics, flow visualization, and evolutionary patterns. Amer. Zool. 40: 101–122. Lindsey, C.C. 1978. Form, function, and locomotory habits in fish. In: Fish Physiology: Locomotion. Vol. VII (W. S. Hoar and D. J. Randall, eds). Academic Press, New York, pp. 1–100. Long, J.H. Jr. B. Adcock and R.G. Root. 2002. Force transmission via axial tendons in undulating fish: a dynamic analysis. Comp. Biochem. Physiol. 133A: 91–929. Long, J.H. Jr, M.J. McHenry and N.C. Boetticher. 1994. Undulatory swimming: How travelling waves are produced and modulated in sunfish (Lepomis gibbosus). J. Exp. Biol. 192: 129–145. Long, J.H. Jr. and K.S. Nipper. 1996. The importance of body stiffness in undulatory propulsion. Am. Zool. 36: 678–694. Long, J.H. Jr. 1998. Muscles, elastic energy, and the dynamics of body stiffness in swimming eels. Am. Zool., 38: 181–202. McHenry, M.J., C.A. Pell and J.H. Jr. Long. 1995. Mechanical control of swimming speed: stiffness and axial wave form in undulating fish models. J. Exp. Biol. 198: 2293–2305. Monod, T. 1968. Le complexe urophore des poissons téléostéens. Mém. Inst. Fond. Afr. Noire, Dakar, 81: 1–705. Nauen, J.C. and G.V. Lauder, G.V. 2002. Hydrodynamics of caudal fin locomotion by chub mackerel, Scomber japonicus (Scombridae). J. Exp. Biol. 205: 1709–1724. Nelson, J.S. 1994. Fishes of the World. 3rd edition. Wiley, New York. Nursall, J.R. 1956. The lateral musculature and the swimming of fish. Proc. Zool. Soc. Lond. 126: 127–143. Rayner, M.D. and M.J. Keenan. 1967. Role of red and white muscles in the swimming of skipjack tuna. Nature 214: 392–393. Rome, L.C., R.P. Funke, R.Mc Neill Alexander, G. Lutz, H. Aldridge, F. Scott and M. Freadman. 1988. Why animals have different muscle fibre types. Nature 335, 824–827. Rome, L.C., D. Swank, D. Corda. 1993. How fish power swimming. Science 261: 340–343.

222

Sánchez-Villagra, M.R., S. Gemballa, S. Nummela, K.K. Smith and W. Maier. 2001. Ontogenetic and phylogenetic transformations of the ear ossicles in mammals. J. Morph. 251: 219–238. Shadwick, R.E., S.L. Katz, K.E. Korsmeyer, T. Knower and J.W. Covell. 1999. Muscle dynamics in skipjack tuna: timing of red muscle shortening in relation to activation and body curvature during steady swimming. J. Exp. Biol. 202: 2139–2150. Shadwick, R.E., J.F. Steffensen, S.L. Katz, and T. Knower. 1998. Muscle dynamics in fish during steady swimming. Am. Zool. 38: 755–770. Shadwick, R.E., H.S. Rapoport and J.M. Fenger. 2002 Structure and function of tuna tail tendons. Comp. Biochem. Physiol. A 133: 1109–1125. Syme, D.A. and R.E. Shadwick. 2002. Effects of longitudinal body position and swimming speed on mechanical power of deep red muscle from skipjack tuna (Katsuwonus pelamis) J. Exp. Biol. 205: 189–200. Videler, J.J. 1993. Fish swimming. London, Chapman & Hall, Fish and Fisheries Series 10. Wainwright, S.A. 1983. To bend a fish. In: Fish Biomechanics (P.W. Webb and D. Weihs, eds). New York: Praeger, pp 86–92. Wainwright, S.A., F. Vosburgh and J.H. Hebrank. 1978. Shark skin. Function in locomotion. Science 202: 747–749. Wardle, C.S. and J.J. Videler. 1993. The timing of the EMG in the lateral myotomes of mackerel and saithe at different swimming speeds. J. Fish Biol. 42: 347–359. Wardle, C.S., J.J. Videler and J.D. Altringham. 1995. Tuning in to fish swimming waves: body form, swimming mode and muscle function. J. Exp. Biol. 198: 1629–1636. Webb, P.W. 1984a. Body form, locomotion, and foraging in aquatic vertebrates. Am. Zool. 24: 107–120. Webb, P.W. 1984b. Form and function in fish swimming. Sci. Am. 251: 58–68. Webb, P.W. 1988. Simple physical principles and vertebrate aquatic locomotion. Am. Zool. 28: 709–725. Weihs, D. 1989. Design features and mechanics of axial locomotion in fish. Am. Zool. 29: 151–160. Westneat, M.W., M.E. Hale, M.J. McHenry and J.H. Jr. Long. 1998. Mechanics of the fast start: muscle function and the role of intramuscular pressure in the escape behavior of Amia calva and Polypterus senegalus. J Exp. Biol. 201:3041–3055. Westneat, M.W., W. Hoese, C.A. Pell and S.A. Wainwright. 1993. The horizontal septum: mechanisms of force transfer in locomotion of scombrid fishes (Scombridae, Perciformes). J. Morph. 217: 183–204. Westneat, M.W. and S.A. Wainwright. 2001. Mechanical design for swimming: muscle, tendon, and bone. In: Fish Physiology 19: Tuna-Physiology, Ecology, and Evolution (B.A. Block, and E.D. Stevens, eds.). Academic Press, San Diego, pp. 272–313.

Zoology 106 (2003) 3