Jb. Anat. 122 av Guytay Fischer Verlag Jena Leh: I fie S) matische 2 gic der Universitit Tabingen (Germany) b On the Anatomy and Development of the Vertebral Column and Pterygiophores in Polypterus seneyaluy Cuvier, 1829 (“Pisces”, Polypteriformes) Perek BaRtscu and Sven Geatmaria With 9 Figures Abstract The ontogeny of the trunk endoskeleton in polypterid fishes is described on the basis of a developmental series from late free embryonic stages onwards. The description focuses on the pattern of axial skeletal elements and the mode of vertebra formation. The bony centra are surprisingly similar to those of teleosts, but preformed by cartilaginous hemicentra. Intercalated arcuals are absent and the 2 series of “basiventrals” of the abdominal region provide basapophyses of dorsal and ventral ribs, respectively. These cartilages, however, are merely transitional structures in early ontogeny; soon they are covered and replaced by bone. Analyses of aquarium observations of larval and adult . Polypterus senegalus were extensively used to characterize their locomotory behaviour and abilities of non-locomotory body movement. A functional interpretation is offered, which emphasizes the unique and specialized nature of the polypterid skeleton, in some aspects paralleled by Lepisosteus. It is concluded that the adaptive role of the polypteriform body skeleton is to allow for a rich variety of exactly guided voluntary movements. Finally, an attempt is presented to align functional and phylogenetic considerations concerning the vertebral column of osteichthyan fishes. A key feature of the polypteriform axial skeleton is the acceleration of vertebral ossification and the concomitant reduction of cartilaginous Stages in the developmental sequence. Usually this feature is considered a characteristic of Teleostei only. 1. Introduction The Polypteriformes, Brachiopterygii or Cladistia consist of 2 living genera: Polypterus, comprising approximately 9 species (Pot. 1965; DaGet and DEs- ourteR 1983; Goss 1984), and the monotypic genus Erpetoichthys Smrrn 1866 _ (Calamoichthys SwxrH 1866). Their distribution is confined to tropical West and Central Africa, inhabiting rivers, lakes, and swamps. They represent a small and Outstanding group of “primitive” osteichthyans. Much scientific interest has ed on their anatomical structure. Complexes of characters which have been studied concern the massive ganoid squamation (KERR 1952; MEUNIER 1980; Sire 1990), the bilobate, lung-like air breathing organ (De SMET 1966; ot and Deswarrines 1967; BrainerD et al. 1989), cranial structure and e (ALus 1922; Moy-THomas 1933; JoLUE 1984; BERING 1991; Zool. Jb, Anat. 122 (1992) 4 497ArRaTIA and SCHULTZE 1991, for example), and the pectoral fin (BRAUs 1901 BupGetr 1901, 1902; Jessen 1972). The pectorals, for example, are equipped with a muscular lobe, their endoskeletal structure is unparalleled in Actinopterygii or “Sarcopterygii”, and functionally they represent an apparatus for slow propulsion There is still some debate about the exact systematic position of polypterid fishes (DaGer 1950; PATTERSON 1982; LAUDER and LiEM 1983; BIERRING 1985Sa, 1991; Arratia and Scuuttze 1991), though Goopricn’s view (1928) of a closer actinopterygian relationship seems to be broadly accepted now (see however BJERRING 1985b). Instead of discussing this phylogenetic problem at length, our main objectives are the details of structure, the development, and functional aspects of the axial and fin skeleton. The polypteriform vertebrae and dorsal fin spines have an unique shape, which enables to assign fragmentary fossil remains from the Cenozoic to this group with good reliability (DaGeT 1950; GreeNwoop 1974). The Polypteriformes are also peculiar in the combination of an elaborate bony vertebral column with a “primitive” dermal armour, which is comparable to that of the extinct palaeonisciform actinopterygians. Among other living fishes, this unusual combination is only found in Lepisosteidae. The precocious judgement that this character combination represents a rare and unusual one, is based on the fossil record, and on the rather vague idea of an evolutionary trend among Osteichthyes towards reduction of the exoskeletal support of the body and its functional replacement by lighter and more “elegant” constructive solutions of the endoskeleton (BREDER 1926; LUND 1967; GUTMANN 1975) However, it should be emphazised that several fossil bony fishes have been reported, showing well developed vertebral centra and massive rhombic scales similar to the condition in Polypterus and Lepisosteus (WrsTOLL 1944; LarrM 1979; SCHULTZE and CHORN 1986). In addition, the supposed functional replacement of the exoskeleton by an ossified endoskeleton is neither realized among “sarcopter- ygian” genera, nor in assumably active and fast swimming fossil Actinopterygii (see BARTSCH [1988] for a survey). Quite often, rather a concomitant reduc- tion of ossifications in both skeletal systems took place, as in the living Dipnoi, Latimeria and Acipenser. We will return to this problem in the dis- cussion. Some accounts on the polypterid axial skeleton are already present in the literature (SEWERTZOFF 1926; ReMANE 1936; Dacet 1950; FRANCOIS 1966b: Bartsch 1988), but neither a complete survey of different body regions, nor a concise functional interpretation has been attempted. The following description is mainly a morphological contribution to the understanding of the evolution and functional specialization of the vertebral column among these fishes, supplemented by aquarium observations of locomotory habits. Polyprerus might provide us with an ontogenetic explanation, as well as an evolutionary scenario: In which way the first steps of centra formation, the expansion of cartilaginous and bony tissue around the notochord can be correlated with the locomotory characteristics a exoskeletal functions in larvae and in “primitive” Osteichthyes respec- tively. 498 Zool. Jb. Anat. 122 (1992) 4 DUsROR GT 2. Materials, methods and some aspects of larval locomotory behaviour To our knowledge, the first breedings of Polypterus in captivity were reported by ARNOULT (1964) and ARMBRUST (1966), In Juli 1988, an attempt at the breeding of Polypterus senegalusin aquaria of the Zoological Institute in K6In was successful. After several fruitless trials with different species and (apparently “egocentric”) individuals, this result can be reproduced now. The essential trigger for reproductive behaviour appears to be the lowering of the conductivity of the water. Like in so many other tropical fresh water fish, the breeding period of polypterids is coupled to the rainy season, when in natural habitats a migration into the flooded areas takes place (BUDGETT 1899; ALBARET 1982). A more detailed description of the reproduction and the embryonic and larval development would deserve a separate publication. For our purpose now it may be sufficient to give a short survey of the stages used in this study The following description of behavioural and locomotory characteristics is based on extensive observations in aquaria and sequences of movements taken by a video camera (strobe-effect shutter function, exposure time: 1 ms) The ciliated eleutheroembryos (definition of the term by BALon [1975]) hatch around the 4thd after egg fertilization (at 25 °C). They stay passive for another 4 to 5d after hatching, attached to plants by glutinous head glands, until the yolk is resorbed and they start active feeding. Axial swimming movements of the free embryos and larvae are anguilliform and confined to short bursts of speed during escape reactions. Most time, the body is held straight and stiff, and the “normal” locomotion of the larvae is performed by a whirling undulation of the pectorals. Slow axial bending, a characteristic orientation behaviour of the adults, is seen more often in a larval stage from approximately 20 mm total length onwards (Fig. 1). Now the “neck” has comparatively large freedom of movement, and voluntary head bending in saggittal and frontal plane is often observed. This is also the size, in which the anguilliform (Lixpsey 1978) undulatory movement of high amplitude and wavelength is gradually replaced by undulation of relatively smaller amplitude and wavelength, confined to the posterior abdominal and caudal region. Lateral body Fig. 1, Three living larval stages of Polyprerus senegalus shown in different postures (approx. 15, 20, and 40. mm total length, drawn from photographs and movie pictures). 35° Zool. Jb. Anat. 122 (1992)4 499excursions during locomotory undulation are thus well beyond the possible maximum amplitude in adult Polypterus (shown in orientation movements, returns, and while creeping over mud and solid ground). It has to be noted, that sustained slow propulsion in adult animals as in larvae is usually performed by the pectorals only. This paraxial “secondary drive” is used for sneaking around and close approach at the prey, for instance Comparatively large pre) start of the active larval period on. Growth is much faster inside the large rearing tank (with a rich supply of different food organisms), than in glass dishes (with Artemia nauplii as a starting food only). However, the death rate of larvae is much higher in the large aquarium, probably because of little control of the food uptake by weaker and smaller individuals. Also considerable ageressiveness was observed among the later larvae, resulting sometimes in severe biting wounds at the tails, pectoral fins, and external gills, Therefore, some care should be taken to avoid misinterpretations of unusual anatomical situations which are due to regenerative processes. (Itis well possible, that this aggressiveness represents a kind of dispersal behaviour, facilitating a wider distribution of the larvae and a freeing from the often only seasonally submerged breeding ground. In contrast to several notes in the aquaristic literature, intra- and interspecific aggression was rarely observed among adult Polypterus.) There is a slow continuous anatomical transition from the larval phase to the juve The external opercular gills can hardly be taken as a landmark of ontogenetic phases, for especially in larger species they may persist into a stage up to 20 or 30cm body length when all other adolescent anatomical characteristics (e.g. squamation, fins, vertebral column) are well differentiated. Furthermore, their development is apparently much influenced by environmental factors The following 6 larval stages and | juvenile were prepared according to a clearing and differential cartilage-bone staining technique (SiIMONs and VAN Horn 1971; DinGeRKus and UHLER 1977): 11.0 mm (13 d), 14.5 mm (15 4), 17.1 mm (20 d), 21.5 mm (27 d), 30.6 mm (40 d), 39.9 mm (40 d), 71.3 mm (approx. 3 months old). The total length of the specimens was taken immediately after death in MS 222 (Sandoz), 1 : 5000, and before fixation in neutral buffered formalin In addition to preparations of adolescent specimens (202 and 214 mm TL), several serially sectioned larvae and small juveniles were available for the analysis can be captured by vigorous suction movement from the Polypterus senegalus CUVIER 1829 22.9 mm TL, § hm transversally, formol, Azan (HEIDENHAIN) — (ISZ), 27.0 mm TL, 7 ym transversally, Bouin, Azan (MALLORY) — (NRS), 34.4 mm TL, 10/12 pm frontally, formol, Azan (HEIENHAIN) — (/SZ) 41.1 mm TL, 8 um transversally, formol, Azan (HEIDENHAIN) — (ISZ). 45.0 mm TL, 8 im transversally, fixation unknown, Azan — (NRS) 70.0 mm TL, 10 jm transversally, formol, Azan (HEIDENHAIN) — (1SZ), 71.0 mm TL, 10 and 12 um frontally, formol, Azan (HEIDENHAIN) — (1SZ), 98.0 mm TL, 10 pm transversally, Bouin, Azan (MALLORY) — (/SZ). Polypterus bichir Georrroy St. HILaiRe 1802 125 mm TL, 20 1m transversally, fix. and staining unknown — (VRS, Expetoichthys calabaricus SMITH 1866 325 mm TL, adult male, transparent preparation, stained for cartilage and bone — (1SZ) From the adolescent specimen of Polypterus senegalus (202 mm TL) and the adult of Erpetoichthys calabaricus (325mm TL), several abdominal and caudal vertebrae were removed, and the associated connective tissue was partly macerated in potassium hydroxide (30%) with some drops of HO; added. The dried vertebrate then were analysed under SEM (Stereoscan 250 Mk 2) to show the fibrous myosept attachment and the con- figuration of the superficial trabecles of bone. 500 Zool. Jb. Anat. 122 (1992) 4 le. From the 34.4mm-specimen, a plate reconstruction of 3 abdominal segments of the vertebral column and associated connective tissues was made in 200-fold magnification. he collection of the Naturhistoriska Riksmuseet in The sections either belos Stockholm (Sweden) — (VRS), or to the Institut fiir Systematische Zoologie, Universitit Tubingen (Germany) ISZ, 2.1. Abbreviations in the figures {ct _actinotrichia Lil mentum intervertebrale inter- in anal fin num Bd _basidorsal LIS lateral line scale c bony centrum LpdC _ \epidotrichia of the caudal fin Cd chorda dorsalis (notochord) Mee membrana elastica externa Che chordocentrum (calcification of Ms ——_myoseptu the notochordal sheet) Mx maxillary Chy ceratohyale Na neural arch Cl cleithrum Nao occipital neural arch Co occipital centrum Nau ural neural arch D dentary Oc otic capsule Fld eminentia ligamenti denticulatum Op —_ operculum (ossified attachment of ligamen- Pan _preanal fin fold tum denticulatum) PHa _ preural haemal arch and spine FRd foramen radicis dorsalis (of spinal (numbering postero-anteriorly) nerve) PHy _ parhypurale (= PHa 1) FRv foramen radicis ventralis (of spinal Prmx _ premaxillary nerve) Prpl__ processus pleuralis Fs fibrous sheet of the notochord Prir processus transversus Ha __haemel arch Psph —_parasphenoid HC (cartilaginous) hemicentrum Pia —_ accessory (distal) radial element of Hmd byomandibula the dorsal fin Hp __ haemal process (infrahaemal) PtAn _pterygiophore of anal fin Hy hypural (numbered from before PD _pterygiophore (radial element) of backwards) the dorsal fin Le labial cartilage Rp! pleural (ventral) rib Ld ligamentum denticulatum Rir transversal (dorsal) rib LH lamina haemalis Sn supraneural Liv ligamentum intervertebrale Spd dorsal spine (externum + internum) Sihy _ stylohyale Lie ligamentum intervertebrale exter- Sir septum transversum (septum hori- num zontale) Ust urostyl Results 3.1. Gross anatomy of the axial and fin skeleton 3.1.1. The sequence of early skeletal development For completeness, the head and pectoral girdle skeleton are shown in the figures too, and will be considered in the description as far as the general developmental pattern and some occipital structures are concerned. In the smallest described specimen (11.0 mm; Fig. 2A), the axial skeleton almost exclusively consists of the notochord, surrounded by a dense fibrous notochordal sheet, the elastic membrane, and a perichordal envelope of connective tissue. Only Zool, Jb. Anat. 122 (1992) 4 501Fig. 2. Skeleton of whole cleared lary: 14.5 mm, (C) 17.1 mm, (D) 21.5 mm; cartila chondral bone shown in black in D), bo: hatching, arrows point to the incisure of the caudal fin fol and dorsal fin skeleton. ae of Polypterus senegalus — (A) 1.0mm, (B) ze stippled (cartilage not covered by peri- ne white, notochord indicated by vertical ld and the division between caudal 502 Zool. Jb. Anat. 122 (1992) 4 few cartilaginous endoskeletal elements are present: the dorso-anteriorly inclined occipital neural arches (Nao), which abut on the posterior face of the otic capsules. In the 3 following neural arches, only the arch bases (basidorsalia) are chondrified At this stage, also the anlage of four hypurals (Hy on Fig. 2A) and the parhypural (PHy) can be detected in the caudal region already (for definition of these elements, see NyBELIN [1963] and Monop [1968]: in most living Actinopterygii, the parhypural is the posteriormost ventral element which surrounds a haemal canal proximally, and marks the bifurcation of the caudal vessels; the position of this bifurcation usually can be taken to delimit ural and preural region of the caudal vertebral column). The hypurals are closely apposed to the elastica externa and the fibrous sheet of the notochord without any apparent arcual contribution. They are situated in the region where the urostylar notochord (US1) bends slightly upwards to reach the dorso-posterior tip of the animal. The parhypural is merely a small infrahaemal knob of cartilage in the smallest specimen, unforked and still separated from the body axis proximally. All other axial skeletal elements of the body are only represented by incipient thickenings of the fibrous connective tissue at the attachment lines of myosepta to the peritoneum, septum horizontale, septum mesiale superius and inferius, and perineural and perichordal sheet, respectively The continuous larval fin fold remains intact up to the 21.5mm stage (Fig. 2D). Actinotrichia (Act) are already visible inside the most caudal part of the earliest larval stage. In this region also the first differentiation of bony lepidotrichia (LpdC) of the caudal fin has started. The latter is confined to‘the hypaxial division of the caudal region (cf. SCHMALHAUSEN 1912, 1913; SEWERTZOFF 1924). In early developmental stages (up to 14.5 mm, Fig. 2B), the caudal fin is only a small ventral outbulging of the larval fin fold, a little resembling the much later differentiating anal fin in front. Soon, however, it assumes a fan-like shape by pronounced distal growth of the lepidotrichia, extending considerably behind the caudal tip of the urostyl (Figs. 1, 2, 3C, 4). In larger juvenile and adult specimens the indentation between dorsal and caudal fin margin (arrow in Fi 3C) disappears along with the relative reduction of urostylar length. This gives the impression of a symmetrical diphycercal condition. The timing and extent of “dorso-caudal fusion” was found to vary much intra- and interspecifically, and can also be influenced by regenerative processes after inflicted damages from attempted cannibalism, for instance. The 14.5 mm larva (Fig. 2B) is interesting with respect to the onset of first perichondral ossification of the Ist and 2nd pair of neural arches. A thin sheet of bone is wrapping the middle part of the arcuals, leaving the base and the apex exposed. This bone is later much extended by membraneous “Zuwachsknochen” (sensu STARCK 1979) or membrane bone (sensu PATTERSON 1977). The exposed basidorsal attachment at the notochordal sheet and the tip of the neural arch may remain some time in ontogeny as growth structures and articulatory areas. This is also true for the apophyses of supraneuruls, infrahaemals, ribs, and pterygio- phores of later stages. The cartilaginous bases of the occipital neural arches in this and the following stage (17.1 mm, Fig. 2C) are connected to the parachordal cartilages anteriorly. Despite massive ossification, these occipital neural arches are 2, Zool. Jb. Anat. 122 (1992) 4 503occiput as separate entities quite a lo Jorsoposterior well discernible from the “primar in ontogeny, and a joint is formed between the neural arch and the ¢ aspect of the otic capsule. Thus, a certain mobility in horizontal and vertical plane ef. ALLIs [1922] for the fused adult is kept during larval development (Fig. 3A; condition). In the 14.5 mm st . the onset of perichondral ossification is also observed in the head skeleton around the ceratohyal and the Ist ceratobranchial element Dermal ossifications obviously start earlier: maxillary, dentary, parasphenoid, re already differentiated in the parts of the ectopierygoid series, and cleithrum a 14.5 mm larva In this stage, the first pterygiophores and rays of the dorsal fin appear in the caudal region. Compared to the younger specimen and the older stages, the notochord arches more strongly upwards caudally, indicating the supposed ancestral situation of a heterocercal caudal fin, This is later concealed by pronounced growth of the dorsal fin and reduction of urostylar length. Whereas the neural arches differentiate in an antero-posterior sequence, the reverse is true of the lepidotrichia and endoskeletal supports of the dorsal fin (Sewerrzorr 1924) Also the hypaxial caudal skeleton shows an anteriorly directed sequence of chondrification and ossification, in line with the late differentiation of the anal fin in the 21.5 mm larva (Fig. 2D). 504 Zool. Jb. Anat. 122 (1992) 4 Fig. 3. Vertebral column and fin skeleton of a cleared and stained larva of Polypierus senegalus 30.6 mm TL (A) axial skeleton of anterior dominal region in lateral view, (B) axial and dorsal fin skeleton of middle abdominal region in dorso-lateral view, (C) caudal skeleton in lateral view; cartilage stippled (fine stippling where covered by perichondral or dermal bone), bone white; small arrow indicates a discontinuity in bony vertebra formation. Two more anatomical characteristics of the skeleton are apparent in the 17.1 mm specimen (Fig. 2C): the occipital region is strengthened by the posterior outgrowths of the parasphenoid /Psph), which closely attach to the occipital centrum in the 21.5 mm stage (see also Fig. 3A). The occipital centrum chondrifies and ossifies well before signs of centra formation are recognizable in more caudal divisions of the axis. Most of the dorsomesial supraneurals are present now. Apparently, these are first strictly metamerically (intermyomerically) arranged and more advanced in growth caudally. Deviations from this regular pattern could only be observed in the anteriormost abdominal region; in the ural region supraneurals are usually absent (Figs. 3, 4). Growth in body length is mainly due to abdominal myomere growth: the muscular caudal peduncle becomes relatively shorter during ontogeny. By pro- nounced growth of the caudal lepidotrichia and despite reduction of the posterior urostyl, the relation between preanal length and total length nevertheless remains quite unchanged around a value of 0.69 to 0.75. The myomere count is almost constant from the larval period onwards (around 50 to 54, with some uncertainities in the occipital and hindmost caudal region). Likewise, the typical Z-shape of the myosepta is already present in the larvae. Only slight changes take place towards the adult condition. Thus, there is no simple intermyomeric relation between axial skeletal elements and myosepta in locomotory active stages. Even the cartilaginous arch bases are strictly confined to the proximal attachment site of the myoseptum only for the first short time during skeletal development. Soon, Zool. Jb. Anat. 122(1992)4 S05the attachment sites of the myosepta obliquely traverse several vertebral segments (see later and Fig. 7) 3.1.2, Characteristics of vertebra and rib formation It is not before approximately 20 mm total length (see the 21.5 mm specimen. Fig. 2D), when the onset of formation of vertebral centra and ribs is observed. Faint constrictions of the notochord are visible now in the anterior abdominal and in the caudal region. These are produced by thin sheets of chondroblasts which are in continuity with the arch bases and spread over the membrana elastica externa of the notochord (Figs. 6, 7). The ossification of centra proceeds in an antero-posterior sequence starting from the postoccipital region, and the vertebral bodies at first have the shape of dorsal half-rings (Figs. 3, 4). The ural centra are the last to differentiate. The development of the vertebral column is accompanied by the formation of scales. First ossifications in the pectoral and caudal region (apparently starting from the lateral line) are observed in the 30.6 mm larva among the cleared and stained specimens. Distinct connective tissue pads inside the corium, however, can already be detected in younger larvae (Fig. 6; cf. MEUNIER et al. [1980], SIRE [1990]), An unusual trait of the polypterid axial skeleton is the existence of 2 pairs of abdominal “basiventral” elements per segment: here named processus transversi (Prir) and processus pleurales (Prpl). They correspond to series of dorsal (transversal) and ventral (pleural) ribs, respectively. Cartilaginous dorsal ribs are already present in the 21.5 mm larva of Polypterus senegalus almost along the entire abdominal length. They are still separated from their prospective proximal articulatory sites, the processus transversi. These are represented in the 21.5 mm specimen as tiny knobs of cartilage, situated inside the septum horizontale of a few of the anteriormost abdominal segments only. In contrast, the few anterior pleural ribs of the 21.5 mm larva are intimately connected to cartilaginous processus pleurales. The latter are much more numerous in the 30.6 mm stage (Fig. 3 A, B), however, not present in all divisions of the abdominal region in each specimen. In the reconstructed 34.4 mm larva, for example, the ventral ribs of the middle and posterior abdomen are still only ligamentously bound to the underface of the transversal processes (Fig. 7). As can be seen in the the posterior abdominal region of the 39.9 mm specimen, formation of cartilaginous ventral rib supports is late and only shortly precedes the ossification of the underface of the centrum (Fig. 4) In general, the proximal joints between ribs and “basiventral” cartilages pers only during larval development. The pleural and transversal cartilages are soon enveloped by perichondral bone and major part of the rib attachment is completed by direct ossification. In later juvenile stages, these “basiventrals” merely remain as small cartilaginous cores of the massive bony processus transversi and centra, and finally become completely reduced. Accordingly, during ontogeny the syn- chondrotic joints (if formed at all) are replaced by connective tissue attachments between ribs and vertebral bone. The proximal limbs of the ribs are hinged to the underface of the bony transversal processes of the differentiated vertebrae in a manner which appears to allow for considerable antero-posterior and rotational 506 Zool. Jb. Anat. 122 (1992) 4 displacement (Prir, Rir in Figs. 5, 6, 8). Furthermore, even the transitional formation ofa synchondrosis between cartilaginous transversal or pleural processes and ribs is confined to anterior abdominal segments of larval stages (Fig. 3A). This is due to the fact, that the proximal end of most ribs is completed by membrane bone, without cartilaginous precursor (Fig. 6A). The mode of rib formation thus bears a strong resemblance to that in Teleostei. The distal limb of most dorsal ribs is thickened, and at least a cartilaginous core remains for a long time in ontogeny. Along the anterior and middle abdominal region it attaches to the inner face of the corresponding lateral line scale by dense connective tissue (Fig. 6A). This distal attachment has quite a variable structure, depending on the developmental stage and the region of the body. In larvae and juveniles, the cartilaginous distal tips in the pectoral region are broadened and strongly bent ventrally and caudally to form a kind of gliding buttressing against the scale (Figs. 3A, 6D). In adults the distal ends are usually received in a distinct internal articulatory groove of the lateral line scale and usually no cartilaginous facets are involved in the formation of the definite joint, neither inside the scale groove nor at the costal apex. Only in | case (a 98 mm juvenile Polypterus senegalus), a fusion of a rib with the scale and formation of a joint fissure inside the distal rib cartilage was observed. The dorsal ribs of adult specimens of Polyprerus and Erpetoichthys are rather short in the posterior abdominal region, and the ossification here not closely approaches the scales. In contrast, the posterior pleural ribs usually grow considerably longer than their anterior serial homologues. The dorsal ribs in adults are not strictly confined to the abdominal region; | or 2 pairs of membrane bones can sometimes Timm Fig. 4. Caudal skeleton of a cleared and stained larva of Polypterus senegalus 39.9 mm TL; cartilage stippled, bone white; small arrows indicate a supernumerary radial element of the dorsal fin, an irregular haemal membrane bone in the anal region, and the articulatory fissure of the preural haemal arches, respectively. Zool. Jb. Anat. 122 (1992) 4 507be found in the anterior caudal region as well. Whereas abdominal and caudal region are separated by the skeletal anatomy in the older larvae (a result of the antero-posterior direction of rib differentiation and the reverse sequence of formation of caudal haemapophyses), the regions rather appear to grade into each other later in ontogeny (Fig. 4). The series of pleural rib supports is now continued into the caudal region as a series of typical haemal arch bases. In the posterior abdominal region also some irregularities can be observed: for example the “precocious” formation of rib- or haemal spine-like membrane bones above and in front of the anal fin skeleton (Figs. 3C and 4, Rp). 3.1.3. The differentiation of caudal and dorsal fin skeleton Before describing the details of vertebra formation in Polypterus, there is need for some further comments on the development of the unpaired fins, The lepidotrichia first emerge as paired continuous laminae of bone within the larval fin fold. They seem to follow bundles of actinotrichia, which appear as precursor structures. The first segmentation of caudal lepidotrichia is visible in the 17.1 mm specimen (Fig. 2B), followed by that of the posterior dorsal lepidotrichia in the 21.5 mm specimen (Fig. 2D). The first distal branching of the upper and lowermost caudal fin rays has taken place in the 30.6 mm larva (Fig. 3C). The characteristic dorsal fin spine formation is just as described by SEWERTZOFF (1924) and DaGET (1950): The proximal paired paris of the anterior rays hypertrophy, fuse, and develop sharp lateral edges as well as dorsally growing paired and pointed tips. The branched and further divided upper segments remain comparatively small Their proximal articulation with the spine is situated at its posterior face and below its apices. Thereby the formerly continuous dorsal fin fold is interrupted and the typical series of dorsal “gaff-sails” is produced. The adult anatomical condition can best be gathered from DaGeT (1950), BJERRING (1985), BARTSCH (1988). The dorsal fin lacks functional mm, inclinatores pinnae, and the fin spine is moved by pairs of mm. erectores et depressores pinnae in sagittal plane only. a Spines and finlets are usually adducted during fast movement. Full erection is regularily associated with surfacing for aerial respiration and other excitements, when the animals are exposed to their potential predators. Sometimes it is also observed during slow axial locomotion. Therefore, we believe it to represent a protective structure in first respect Nevertheless, the hydrodynamic properties of such a fin might be adjustable and similar to a continuous dorsal fin with lateral flexibility, owing to the “windows” opened up between the more or less erected finlets. The proximal articulatory part of the fin spines embrace a thickened distal cartilaginous knob of the pterygiophores. The latter are not metamerically arranged, but are almost horizontally positioned and always attached to the dorsal lepidotrichia in a regular one to one relationship. In the posterior abdominal and anterior caudal region, they may reach between the supraneurals of the vertebral column proximally. Secondarily, small cartilaginous dorsal pterygiophore elements develop caudally of the main articulation (Figs. 3C, 4). They vary a little in shape and position in early ontogeny (see the supernumerary cartilage in the 39.9 mm specimen, Fig. 4), but are usually associated with the site of origin of the erector 508 Zool. Jb. Anat. 122 (1992) 4 muscles of the finlet next behind. They are smaller and differentiate later anteriorly, on. 's considerably from this specialized or can even be absent in the anterior abdominal and posterior caudal r The endoskeleton of the anal fin di arrangement. It simply consists of 2 series of closely gathered and sometimes fused radial elements. In juvenile and adult specimens, the anterior distal radials show an extraordinarily large rounded articulatory knob yentrally. The anal fin of adults, especially the enlarged anal fin of males, is equipped with a massive inclinator musculature, producing the well known scale covered muscular lobe. Lateral bending of this fin seems to have an important f behaviour (ARMBRUST 1966; pers. obsery.), probably keeping for fertilization. ction during courtship and sperms close together The proximal bases of the caudal lepidotrichia are covering the lateral aspect of the hypurals and preural haemal spines (Figs. 3C, 4), a condition which provides skeletal stiffening of the caudal fin joint. The pointed proximal tips of these basal fin ray elements are ligamentously anchored to the caudal myosepts and the thick connective tissue of the corium. A similar overlap can be found inside the upper caudal fin of Lepisosteus, for example (cf. NyBELIN 1977). Ventral caudal radials, however, which are separate endoskeletal elements interposed between hypurals and caudal lepidotrichia in Acipenser, Lepisosteus, and a number of “primitive” fossil actinopterygians, are absent in polypterids. There is very little differentiated caudal musculature in Polypterus, except some small mm. flexores yentrales, originating from the ventral caudal myosepts and attached to the elongate bases of the fin rays. A distinct so-called hypochordal longitudinal muscle, as described by SCHMALHAUSEN (1913) was not found. Accordingly, the caudal fin is moved as one entity, with little ability of adjustments and control of individual rays. The branching of the caudal blood vessels, which quite exactly defines the preural and ural region in Amiidae and teleosts (cf. SCHULTZE and ARRATIA 1986, 1988), is often asymmetrical and variable in polypterids (as is the case to an even greater extent in Acipenser and Lepisosteus). Quite often the parhypural is also irregular proximally, and does not always show complete and symmetrical haemal arches (PHy, Figs. 3C, 4); sometimes the haemal spine is absent. Accordingly, the number of ural centra and hypurals varies in adult Polyprerus senegalus (usually 4 to 5 centra and 5 to 6 hypurals). Whereas the preural haemal arches and hypurals have a rather simple attachment at the ventral aspect of the notochord in the younger larvae, an articulatory fissure has appeared in late larvae and juveniles (Figs. 4, 5). The onset of division of the cartilaginous preural haemal arch bases can be detected in the 39.9mm specimen (Fig. 4, small arrow). The mobile articulation of the hypurals and praeural haemal arches is retained during later development of the caudal vertebral bodies. Thus, in juvenile and adult specimens, 2joints are responsible for caudal flexibility against movements of the axial skeleton. This again is similar to the situation in Lepisosieus. In contrast, the hypurals of Amia and the preural haemal arches of many teleosts show primary continuity or secondary fusion with ventral caudal arcocentra (cf. ScHuLTzE and ArRATIA 1986, 1989). The posterior end of the urostyl was always observed to chondrify in adolescent specimens. Zool. Jb. Anat. 122 (1992) 4 509A detailed account on the highly interesting anatomy and development of the paired appendages is beyond the scope of this paper. For a valuable contribution to the anatomy and further references on this subject, the reader may be referred to the publ ion of Jessen (1972). Only a few remarks on its endoskeletal structure may be useful to include here for understanding of the pattern shown in the figures. The endoskeleton of the pectoral extremity of the younger described larvae (Fig. 2A, B) includes a massive cartilaginous scapulocoracoid proximally, at- tached to the posterior face of the cleithrum. Distally, a large cartilaginous plate is connected to the rounded articulatory knob of the scapulocoracoid. This flat and continuous plate is equipped with thickened fringes only, and represents the endoskeleton of the fin. It is not before the 21.5 mm stage (Fig. 2D) when the first series of distal radials is formed in continuity with the axial basal plate. The Just mentioned lateral fringes are now differentiated into separate prae- and postaxial basal radials, embracing the middle plate in the characteristic manner They are the first endoskeletal elements of the pectoral fin to ossify (Fig. 5). The pelvic fins and their supporting endoskeleton was also first detected in the 21.5 mm specimen Fig. 5. Polypierus senegalus, juvenile, 71.3 mm TL, drawn after a transparent preparation; cartilage stippled, bone outlined only. 3.2. Development of vertebral centra As mentioned before, and as can be gathered from the description above, there is no trace of intercalary elements of the arcual series in Polypterus. The caudal haemapophyses are secondarily divided basiventrals. Among living Osteichthyes, only in Chondrostei, Neoceratodus and Amia interdorsals and interventrals are interposed in a regular manner between the basidorsals and basiventrals respec- tively. In Lepisosteus, a few interdorsals can be observed in the ural region. In those osteichthyan groups in which intercalaries have been described, they usually show a greater variability in shape and distribution than the definite arch bases, and often remain cartilaginous. The processus transversi and processus pleurales in polypterids present a difficult problem of homologization with the hypothetical ancestral arch elements (Gapow und Axpor 1895). They form in situ and cannot be derived from either basidorsals or basiventrals ontogenetically. These rib supports are always well separated from each other: a common origin from a theoretically uniform haemal arch base clement is therefore not traceable. This is quite different from the 510 Zool. Jb. Anat. 122 (1992) 4 lll |e ee ere situation in Amtia ealva, in which also 2 “basiventral” series are present along the abdominal notochord: dorsally displaced basapophyses of pleural (!) ribs and separate haemal supports ventrally grading into one entity posteriorly (SCHAUINS- LAND 1906). Their site of formation in Polypierus rather appears to be ruled by Zool. Jb. Anat. 122 (1992) 4 su512 Zool. Jb. Anat. 122 (1992) 4 the attachment site of the ribs only: the line of section between septum horizontale and myoseptum, and between peritoneum and myoseptum, respectively. Further support for this idea is provided by the fact, that processus transversi and processus pleurales form much later than the basidorsal elements and along with rib differentiation, In addition, their position is subject to some variation relative to each other and the basidorsal. The pleural rib supportis often found ina transversal plane anterior to that of basidorsal and transversal process (Fig. 3) The peculiar mode of cartilaginous preformation of the bony centra has already been mentioned by FRANCOIS (1966b) and BartscH (1988). The developing perichordal sheet of chondroblasts (HC in Figs. 6B, C and 7), which precedes bony centra formation, establishes continuity between the arch- and rib bases, apparently spreading out from the basidorsals. Therefore it can be interpreted as “reminiscent” of an arcocentric vertebra formation, comparable to the situation in Lepisosteus (SCHULTZE and ARRATIA 1986). In the middle region of the later vertebral body in Polypterus, an ossification constricts and later replaces the continuous, or already segmented cellular sheet (see Figs. 2D, 3C, 6D, E). This proliferating tissue remains underneath the developing bone in the vicinity of the intervertebral region only. Itis thus very similar to the cellular layers, SCHULTZE and ARRaTIA (1988) described in ural and preural centra of Elops saurus, and may represent a persisting growth structure On the other hand, like a teleostean or amioid “autocentrum” (cf. ScHAUINs LaND 1906, REMANE 1936), the bony vertebral centrum has not always an exact topographical relationship to the arch and rib bases (see the irregularities in Figs. 3A, B). The bone also invades the margins of the attachment sites of the basidorsal cartilages at the notochordal sheet (Fig, 6E, F). The perichondral central ossification first appears as a dorsal half ring, successively spreading over the lateral aspect of the cartilaginous hemicentrum. In younger stages, it still leaves the arch bases and rib supports partly exposed (Figs. 4, 7) There is a typically arcocentric contribution to the vertebral bone formation, which is readily visible in the irregular part of the abdominal region of the 30.6 mm specimen (Fig. 3B). Here, the bone lamina, which covers part of the neural arches has spread out, failing to meet the displaced “autocentrum”, ventrally. Instead, it forms only an independent bony connection of the pair of basidorsals underneath the neural tube, the floor of the neural canal (see also Fig, 6C). In this place, also the attachment sites of the thickened strands of the meningeal membrane (Fig. 6D, E: Ld, ligamentum denticulatum (ef, BJERRING 1991)) during later stages ossify in the form of longitudinal ridges or eminences (Eld in Fig. 9B: Bartscu < Fig. 6. Microscopical sections of the axial skeleton of larval Polypterus. A: P. senegalus, 34.4mm, middle abdominal region, horizontally, B: P. senegalus, 45.0 mm, middle ab- dominal region, transversally, C: P. senegalus, 41.1 mm, middle abdominal region, transver- sally, D: P. senegalus, 98.0 mm, posterior abdominal region, cross section in a plane close to intervertebral ligament, E: P. senegalus, 98.0 mm, anteriormost caudal region, mid- vertebral cross section, F: P. bichir, 125.0 mm, anterior abdominal region, cross section, cancellous bone of the vertebral body (scale bar corresponds to 100 jm). 36 Zool. Jb. Anat. 122 (1992) 4 513[1988]. The deseribed irregularity is certainly not a localized damage, but the central ossifications can be seen to fall more and more out of pace in relation to the neural arches over several segments. The first ossifications of the ver tebral column were also traced in the serial sectioned and reconstructed specimen (Fig. 7). From this model, and the sections of a 41.1 mm larva (Fig. 6C) it is apparent, that there are in fact no vertebral ossifications independent of the cartilaginous hemicentrum and the “basal cartilages” up to these stages. The anterior and posterior margins of the hemicentrum and the periphery of the attachment sites of basidorsals and “basiventrals” ossify first. The extensive contribution of membrane bone to the definite vertebral body has to be considered as a later addition. The reconstruction also shows, that vertebral centra in early development are by no means formed exactly intermyomerically, but for the larger part intramyomerically in Polypterus. Judging from the survey of teleosts, given by RAMANUJAM (1929), FARUQI (1935), and FRANCOIS (1966a), this is not unusual Fig. 7. Reconstruction of the axial skeleton of the middle abdominal region of a larval Polypterus senegalus (34.4 mm TL) in oblique postero-lateral view. Drawn from a plate model x 200. Bony covering of the posterior centrum and arches removed and outline of the chordacentra (notochordal calcifications) shown by interrupted line. Attachment of myosepta simplified 514 Zool. Jb. Anat, 122 (1992) 4 The myoseptum is mainly attached at the front edge of the developing centrum, and there seems to be nothing like an embryonic “resegmentation” of sclerotomic material among actinopterygian fishes (SCHAEFFER 1967; WILLIAMS 1959), (“Re- segmentation” can either be brought about by a separation of sclerotomal halves and displacement towards 2 successive myomeres in some tetrapods in early onto- geny, or by a fusion of already formed bony half centra in some Actinopterygii (REMANe 1936).) The interconnection of body musculature and the axial skeleton already in larval Polypierus is provided by series of 3 successive neural arches, which are obliquely traversed by the attachment site of the same myoseptum (Fig. 7, Ms). In a similar manner as the long neural spines of Amia and teleosts, the outgrowing supraneurals of Polypterus always pass over sevel al muscular segments distally, leaving only the proximal articulatory site in the original segmental position. The supraneurals remain moyeable with respect to the neural arches in juveniles and adults (Figs. 5, 8), and are inclined at a low angle of 15 against the horizontal line. No formation of neural spines by outgrowth of neural arches occurs. A little carelessly, we used the convenient term “attachment” for the description of the relation between myosepia and vertebral ossifications above. From the SEM-pictures (Fig. 9A —D) it is, however, quite clear that there is no distinct site of origin of myoseptal fibres at central structures in the adult Only the prominent dorsal and horizontal ridges of the transversal process provide such direct attachment lines of myosepta and septum horizontale, respectively. The central vertebral body instead, is deeply embedded into a continous dense sheet of perichordal connective tissue. A mass of collagenous fibres is entering the bone of the centrum at the attachment site of the external intervertebral ligament only. Formation of so-called chordocentra, half ring-shaped acellular calcifications of the fibrous notochordal sheet, also takes place in advanced larvae of Polyprerus (Che in Figs. 6, 7, 8). However, they do not precede vertebral ossification, as is the case in many teleosts and Amia calva (cf. RAMANUJAM 1929: FRANCOIS 19662: SCHULTZE and ARRATIA 1986, 1988; BARTSCH 1988). These calcifications probably are a secondary effect of the immobilization of the collagenous fibre system. In adult specimens, they form the inner part of the constricted middle vertebral region (Figs. 8, 9D). In view of the absence of any cellular invasion underneath the outer elastic membrane, they should not be confused with cellular chordacentric formations in Chondrichthyes and Dipnoi, though RaMANUJAM (1929) suggested such a homology. In adults only a short intervertebral distance is bridged by internal and external intervertebral ligaments, which are derivatives of the notochordal sheet and the perichordal connective tissue, respectively (TRETIAKOFF 1926; FRANCOIS 1966a). The vacuolate inner notochordal tissue fills the large space between the concave internal faces of subsequent vertebral bicones as a kind of “nucleus pulposus” in juveniles and adults. As in teleosts, the notochord apparently still performs the function of keeping the length constancy and pressure resistancy of the axial skeleton inside the intervertebral joint (see also TRETIAKOFF 1926; Symmons 1979). 36* Zool. Jb. Anat. 122 (1992) 4 515Therefore it has to be considered as an important functional element of the piscine vertebral column even in the advanced ontogenetic stages The cartilaginous neural arches of either side are closely apposed dorsally and in anterior parts of the vertebral column, their distal ends fuse with each other midsagittally. In this place they produce a massive cartilage, which becomes isolated from the remainder of the cartilaginous neural arches in juveniles, and forms the characteristic joint with the supraneural (Fig. 6E). In its final shape, this becomes a peculiar rotational joint, and the cartilaginous investment obliterates in juveniles of larger species and adults of Polyprerus senegalus (Fig. 8). In addition, in front of this articulation, the roof of the neural canal is closed by fusion of neural arch bones, sometimes only by syndesmotic sutures, in the dorsomesial line. As in living Dipnoi, there is no dorsal longitudinal ligament, which otherwise is a quite constant and functionally important component of the vertebral column in lower vertebrates (in contrast to the notion of Symmons [1979], who considered this dorsal ligament to be present in all vertebrates) Suprainterdorsals are small separate cartilages (in Lepisosteus, Amia, and Teleo- stei), or dorsomesial cartilaginous processes of the neural arches (in Chon- drostei). In these fishes, they are situated above the neural canal and underneath the ligamentum longitudinale dorsale (SCHAUINSLAND 1906, BARTSCH 1988). These elements are absent in polypterids, unless we want to consider them homologous to the above mentioned transitional cartilaginous articulatory complex inside the Fig. 8. Vertebra of an adult specimen of Polypierus senegalus, 202 mm TL (from BARTSCH 1988); oblique anterior view on sagittal cutting surface. 516 Zool. Jb. Anat. 122 (1992) 4 EP Fig. 9. Scanning electronmicroscopic documentation of posterior abdominal vertebrae of Expetoichihys calabaricus (325 mm. TL); scale bar corresponds to 400 um. A: survey in antero-dorso-lateral view, showing the anteriorly inclined transversal process, B: the same aspect as in A, details of the bony surface and attachment site of the external intervertebral ligament, C: lateral view on vertebral body, dorsal strutting of the transversal process and foramina of spinal nerves, D: anterior view on the vertebral edge, and into the vertebral cone: note the rugosities of the intervertebral ligamentous attachment (Lie), E: antero-dorsal view into the neural canal; note the attachment eminences for the denticulate ligament (£Ld), F: ventral view on the vertebral body, showing the haemal laminae and underfaces of the transversal processes Zool. Jb. Anat. 122 (1992) 4 S17roof of the neural canal in juvenile Polypterus. Only in a general sense, both are isolated cartilaginous differentiations of the upper ends of the neural arches The vertebral bicones and neural arches in Polypterus are soon completed and stiffened by lamellae of membraneous bone, as was described for teleosts by LaERM (1976). In contrast to teleosts, there are few centrifugally growing trabecles of bone in polypterids, and the contact between newly formed membrane bone and the primary perichondral ossification of arch bases and hemicentra is first kept (Figs. 6B, F). This secondary process of stiffening seems to take place in polypterids more slowly than in teleosts, mainly by concentric peripheral growth of bone. A pronounced development of bony struts occurs at the lateral face of the centrum in the region of the transversal process, at the ventral aspect by haemal laminae (LH in Figs. 6E, 8, 9F), and dorsally by the antero-posteriorly extended neural arches (Figs. 7, 8). The latter secondarily surround the exits of the spinal nerve roots in a somewhat irregular manner: usually, the anterior ventral (motoric) root of a spinal nerve is enclosed by the posterior bony lamella of the neural arch. The posterior dorsal (sensory) root in some c foramen inside the anterior neural arch bone of the vertebra next behind, or leaves the neural canal between 2 successive neural arches (FRd, FRv in Figs. 7, 8,9). In a few of the most anterior vertebrae, dorsal and ventral root can also take their course through foramina of the neural arch at approximately the same transverse level. s exits from a 4. Summary and conclusions The axial skeleton as documented here in Polypierus senegalus is far advanced compared to the assumed ancestral condition of bony fishes. Many elements (vertebral centra, neural arches, ribs) are completed by membrane bone, the cartilaginous precursors being more or less reduced. Even joints formed in early ontogeny between cartilaginous facets, usually become replaced by ligamentous articulations between membrane bones. Of some general interest is the conclusion that the expected pressure and shearing forces are not necessarily received by cartilaginous tissue, but in the fully differentiated state quite often by systems of fibrous suspensions of tensile strength alone. This can be exemplified by the peculiar attachment joint of the dorsal ribs at the inner face of the lateral line scales, the proximal rib attachment, and the articulation of the supraneural. An exception to this rule are the cartilaginous articulatory heads of the pterygiophores and haemal spines. The fact that the transition from one “constructional type” of endoskeleton towards another can be shown in the ontogeny of polypterids, has an important theoretical impact. A statement: “The ribs are formed entirely independently of the parapophyses and only become secondarily attached to the parapophyses” (FARUQI 1935, p. 332) expresses a wrong view on the function and development of the skeleton, even in teleosts. Though the authors are certainly not free of the prejudice to accept “concrete bones” as separate phylogenetic entities, we will try to avoid the complete loss of connective “soft tissue” beyond the table. 518 Zool. Jb. Anat. 122 (1992) 4 The definite vertebral bodies are similar to those of teleosts. The deeply amphicoelic vertebrae are supported externally by bars of massive cancellous bone, and the cartilaginous arches and rib bases are soon constricted to relictant cores within the bony covering. These arches and rib bases do not well match the schematic concept of regular basidorsal and interdorsal — basiventral and interventral cartilages of Gapow and Appot’s (1895) nomenclature. Intercalaries are interpreted as “more directly functionally influenced” parts of little systematic significance (REMANE 1936; SCHAEFFER 1967; LAERM 1979). This opinion is perhaps better expressed by the note, that the function of intercalated intramyomeric pressure members around the neural and haemal canal becomes redundant with more elaborate ossification of extremities and/or centra of the axial skeleton. Nevertheless, the described 2 series of basapophyses together could represent a divided ancestral basiventral element. But, there is little ontogenetic evidence in favour of such idealized interpretation in our animals. The overall impression is that the vertebral column of polypterids is ruled by strict functional demand rather than by the “ballast” of evolutionary history The absence of the dorsal longitudinal ligament in Polypterus correlates to the specialization of the supraneurals, which form a series of dorsal rods, inclined at a low angle within the sagittal plane and tightly bound together by fibrous tissue. Thereby they establish a certain mechanical continuity over the vertebral segments. However, some elastic recoil mechanism of the axial skeleton, as proposed by Symmons (1979) for the function of the dorsal longitudinal ligament, seems improbable in this case. Certainly, there is need for more experimental investiga- tions until the mobility and functions of all these components of the axial skeleton are clarified. Some preliminary functional conclusions, however, can be gathered already from the ontogeny and anatomical construction of the axial skeleton. The centra formation of Amia calva and Teleostei is usually characterized by the occurrence of independent direct perichordal ossification (autocentra), enclos- ing bone laminae, which grow out from the arcual bases (arcocentra, see SCHULTZE and ArRaTIA[1986, 1988]). The “slow” and extensive spreading of the cartilaginous arch bases around the notochord in Acipenser and Lepisosteus, on the other hand, gives an impression of the more primitive, purely arcocentric process. (Lepisosteus, as is well known, is specialized with respect to the formation of a cartilaginous intervertebral disc and a highly moveable opisthocoelic intervertebral joint.) However, the only major difference between the 2 ontogenetic modes and evolutionary grades is the reduction of cartilage as a precursor of vertebral bone formation. Judging from comparison with the axial anatomy and histology in Chondrichthyes (ScHauinsLanp 1906; RotH 1911; WuRMBACH 1932), Chon- drostei (SCHAUINSLAND 1906), Dipnoi (MooxErsEE 1954; SCHULTZE 1970; BARTSCH 1989), Rhipidistia (ANDREWs and WesTOLL 1970 a, b), and Tetrapoda (WILLIAMS 1959), the ability of arcocentric (and chordacentric) chondrification of the notochordal axis is primitive among Osteichthyes. The direct ossification of perichordal connective tissue in Amia and teleosts is a derived character state of actinopterygians, and a significant phylogenetic trend (EMELIANov 1939, 1940). It accelerates the development of vertebral centra and the functional replacement Zool. Jb. Anat. 122 (1992) 4 S19of a complete notochord. Polyprerus represents an intermediate functional stage in this respect, because there is an incipient cartilaginous preformation of centra soon overruled by growth of arco- and autocentric bone. Further reduction of this perichordal cartilage would exactly result in the “typical” teleostean condition Therefore autocentra are not completely new acquisitions, but probably they have to be derived from cartilaginous and bony areocentra in phylogeny by abbreviation of the developmental sequence. . The implicit idea of a different genetic origin and separate anlage of the teleostean autocentra was corroborated by the independent start of ossification and slight relative positional variations with respect to the arcuals (REMANE 1936). Such irregularities are sometimes observed in Polypterus as well, which justifies the application of the term “autocentrum’” despite its cartilaginous preformation. The possibility should be kept in mind that among fishes, the exact position of the intervertebral joints is not necessarily realized arbitrarily by a detailed morphogenetic pattern, but superimposed by the actual muscular forces on a continuous potentially skeletogenous envelope of the notochord. In the description, we emphasized the early completion of the caudal and posterior dorsal fin skeleton. This pattern is not surprising, since the posterior abdominal and the caudal region provide the main locomotory apparatus for undulatory movements. The developmental sequence, which seems to be basically similar in all actinopterygian fishes, is expected to represent an ancestral condition. However, with the exception of certain ural and preural centra of advanced teleosts (PorrHorr 1974, 1975, for example), formation of caudal bony vertebrae takes place rather late in ontogeny. This indicates that central ossifications appear first in regions, where little or no regularily repeated bending and torsion movements are excerted upon the axial connective tissue. The occipital and pectoral region generally is the most stable one in fishes with respect to such movements. Furthermore, it has to be assumed that the pressure forces of undulatory propulsion in fish are centered on the occipital vertebral column and cranium, which cor- responds to the early and massive chondrification of the occipital centra and/or arches. An extreme immobilization of this region is observed in living Chondrostei by perichordal fusion of several vertebral segments; a condition which is well in line with the stable head posture of these fishes. The dorso-anterior inclination of the occipital neural arches is a common character of osteichthyan fishes, though seldom is such a significant occipital buttressing system produced as in Polypterus- larvae. The distinct gap between the Ist and 2nd pair of neural arches allows for elevation of the head during suction feeding and the mentioned slow dorsal neck bending. The general histogenetic and mechanical approach at the characteristics of skeletal tissue formation (Pauwers 1965) might also explain, why the onset of central ossification is confined to areas, where tensile forces are transmitted from the myosepta to the perichordal connective tissue (compare inter alia FARUQI [1935] for Gadus). In early larval stages, these areas are already stable points and protected from pressure deformation by cartilaginous arcuals, or additionally by cartilaginous hemicentra in Polypterus. These hemicentra do not form exactly 520 Zool. Jb. Anat. 122 (1992) 4 CB T:COC within the attachment sites of the myosepta at the perichordal and perineural sheet, but soon extend far beyond these attachments. The intramyomerie arrange- ment of cartilage in early ontogeny should correspond to sites of alternating pressure forces. The anterior and posterior margins of these cartilages, which perichondrally ossify first, are situated in line with the trajectory of the myoseptal fiber attachment close to the intervertebral joint, instead. This early pattern of ossification in Polypterus is also similar to Teleostei and somewhat reminiscent of diplospondylic concrescent centra present in Amia. However, because of the teleologic intention of this view, we are in doubt if “resegmentation” by inter- myomeric fusion of half centra is tenable as a general explanatory concept for vertebra formation. This was proposed by SCHAUINSLAND (1906) and REMANE (1936), who took some fusing half centra and arches in Salmo as an example In addition to developmental accidents, like somitic irregularities, the extended cartilaginous cover of the notochord provides a mechanical explanation, why Vertebral structures can fall out of pace, as was described in the abdominal region of the 30.6 mm specimen of Polypierus senegalus (Fig. 3B). In the 21.5 mm larva (Fig. 2D), faint divisions of the (“procartilaginous”) perichordal tissue are visible in the anterior abdominal and in the caudal region. These are also the regions, where strongest bending occurs in these advanced larval stages, the typical head-bending and the largest undulatory amplitude respectively Fusion of centra, crescent shaped dorsal and ventral half centra, and relative displacements of neural and haemal arches are by no means rare events among bony fish, despite a regular segmental arrangement in general. Especially Amia calva, showing retained intercalary arcual elements and concrescent half centra, is famous for such “disorderly engineering” in its basically diplospondylic preural vertebral column (SCHAUINSLAND 1906; REMANE 1936). Considerably less variation is found among teleosts. It is to be expected, that these variations are more rare the stronger the axial skeleton is subject to regular repetitive influences of muscular forces. The same trend is to be expected, the earlier and the more complete the continuously flexible notochord is replaced (or constricted in growth in the vertebral region) by a limited number of distinct skeletal elements. The individual segment of the chain then becomes much more important in functional larval stages. The critical point is, to overcome possible pressure and torsion strains within the muscular segment, when the first faint centra start to constrict segments of the pressure stable and continuously flexible notochord in ontogeny. There appear to exist several “convergent solutions” to this problem among fishes, which can be exemplified by diplo- and polyspondylic constructions among Rhipidistia, Chondrichthyes, and halecostome actinopterygians. Functionally, these have to be interpreted as a mean for enlargement of the number of joints, thereby reducing the bending angle of the intervertebral articulation and the load upon the individual central skeletal element. Independently in many groups, especially in rather large sized, but in sluggish as well as in locomotory active animals, we also quite often find a retention of, or a “return” to the archaic notochordal axis, equipped with a thick fibrous sheet (Latimeria, living Chondrostei, and the fossil Pachycormidae, for example). This indicates, that for aquatic propulsion the “backbone” is not Zool. Jb. Anat. 122 (1992) 4 S21necessarily the superior constructive solution. Instead, only under certain cireum- stances of developmental and locomotory adaptations, a replacement of the notochord and its fibrous sheet is possible, without inflicting axial swimming performances. Thus, the onset of deposition of central axial hard tissues in phylogeny must be interpreted as a rather passive process, interdependent with restrictions in mobility and appearance of wearless zones within the chordal and perichordal connective tissue, before the new structures can be “aimed” at a new function. 2 An experimental design seems within reach to answer some of the presented questions, but this approach can only be very limited with respect to the forces imposed on the axial skeleton at different developmental stages. Especially. it is very difficult to evaluate the influence of animal size and of the absolute value of muscular forces. Therefore, we are mainly confined to natures experiment and the comparative method. : The hindmost ural region of the notochord in juvenile and adult stages is covered by continuous cartilage and associated with irregular, sometimes confluent areuals in Polypterus as in many other fishes (see Ust, Nau in Figs. 3C, 4; compare also the variation of ural arcuals in Acipenser and Lepisosteus [BARTSCH 1988) In this place (as at the anterior tip of the notochord embedded into the cranial base) also an arco- and chordacentric process of chondrification often occurs in Osteichthyes. This terminal, urostylar part of the caudal notochord is usually not exposed to strong direct forces of the metameric musculature. Instead, if moveable at all, in most fishes it only follows caudal bending passively, or is moved as one stiff entity. Thus, it might provide a picture of vertebral development without the regulating influence of metameric muscle action. Only in few specialized and quite different cases, a regular (but not necessarily always metameric) segmentation of this structure is kept (see ANDREWS [1976] for Latimeria, BARTSCH [1989] for Protopterus, KiRSCHBAUM and MEUNIER [1981] for the regenerative urostyl of gymnotiform teleosts). We have to accept the fact, that there are different modes and evolutionary stages of vertebra formation present along the same vertebral column in different regions. The vertebral segments and fins are of course only serially homologous. and therefore it is important to give a survey over the complete length for characterization (compare SCHULTZE and ARRATIA [1989] for teleosts) As we indicated in the introduction, neither the exact function of an ossified axial skeleton, nor that of a complicated armour of oblique pressure resisting scale rows has been clarified yet. Likewise, some sort of mutual interdependence of these 2 skeletal systems in phylogeny is obscure. There is thus much need for a refinement of our functional and phylogenetic ideas about the main locomotory apparatus of fish. We made a motion analysis to characterize the ontogenetic change and exoskeletal prerequisites of axial propulsion in Polyprerus (GEMBALLA and Bartscu in prep.), and some of the results shall be introduced here already. In contrast to former ideas (LUND 1967, for example), the squamation does not represent anything like a bending-restrictive device. Rather the opposite is true, because there are few living fishes with a comparable vertebra count which are capable of such narrow bending radius, as polypterids are. Manipulation of $22 Zool. Jb. Anat. 122 (1992) 4 anaesthesized specimens and of various skeleton and connective tissue preparations could neither corroborate such exoskeletal bending restriction, nor unequivocally demonstrate a torsion hindering mechanism (in contrast to the opinion of GUTMANN [1975], PEARSON [1981]. The dorsal ribs of Polypteriformes have been interpreted as lateral struts of the axial skeleton against the dermal armour (SEWERTZoFF 1926), or as facilities for “immobilization of the metameric muscle trajectories” (notion of PEARSON 1981). Both could relate to the remarkable ability of strong lateral bending or straight guiding of the abdominal region, which can be observed in Polypierus and Erpetoichthys, independent of locomotory undula- tion. The large processus transversi at least, are indeed very suggestive as lever-arms of the metameric musculature (Fig. 9). In addition, from the analysis of motion pictures of swimming adult polypterids, it seems that they are perfectly able to damp the reactive forces of caudal undulation upon the anterior body. In contrast to most other fishes, there is no lateral swing of the head out of the straight line of translation during sustained fast swimming. From the ontogenetic stage of vertebra and scale formation onwards, the undulatory movement is effectively confined to one wavelength extending over the posterior abdominal and the caudal region only. For a “primitive” fish which in addition is by no means specialized for high sustained swimming speeds, this is a surprising observation. The polypterid body skeleton also can be characterized by the relative shortening of the caudal region and the lengthening of the abdomen; this process, which is stretched to its extreme in Erpetoichthys, leaves the tail as a kind of pedunculate paddle. This “eel-like” genus is well able to move in an anguilliform manner, but during slow locomotion a body undulation is often uncoupled from the caudal “wrigging”. In conclusion, the discrete elements of the axial skeleton and dermal armour of polypterids rather serve the function of refinement of muscular control Apart from the possible protective function of the exoskeleton, together these systems form a highly specialized locomotory apparatus, which permits an extraordinary variety of movement patterns. In view of the richly structured and seasonally changing environment, this has to be considered as a major adaptive goal at least. Finally, to sum up what we have now of phylogenetic evaluation of characters in Polypteriformes, we provide the following list, which is an impressive mixture of very “primitive” and “advanced” features: 1. Absence of cartilaginous chordacentra over the “free” length of vertebral column; acellular calcareous chordacentra instead — derived osteichthyan character. Absence of intercalaries — derived osteichthyan. Perichordal cartilaginous hemicentra — derived osteichthyan. Arcocentric ossification — probably derived osteichthyan. Autocentric ossification — derived osteichthyan. The combination of the former points: differentiated deeply biconcave mono- spondylous centra, consisting of arco- and autocentric ossifications, and noto- chordal calcifications — derived actinopterygian? DUwswN Zool. Jb. Anat. 122 (1992) 4 523,