Mitt. Mus. Nat.kd. Berl., Geowiss. Reihe l(1998) 81-92 19.11.1998
Conodonts, Calcichordates and the Origin of Vertebrates
Jan Bergstrom ’, Wilfried W. Naumann 2, Jens Viehweg & M6nica Marti-Mus
With 6 Figures
Abstract
Interpretation of early deuterostome evolution and relationships has been hampered by the lack of soft-part preservation in most groups. In addition, a recently revealed upside-down life orientation of vertebrates (the only real notoneuralians) compa- red to other bilateral animals has been misinterpreted as evidence for a unique body design in all deuterostomes, misleading any search for relatives. Regarding echinoderms, the variety of body plans is confusing. The interpretation of some fossils with echinoderm-type calcite skeletons as “calcichordate” ancestors of chordates, however, involves a hypothetical reconstruction of an unusual body plan and a long series of hypothetical transitions. The number of necessary steps is much lower if cephalo- chordates (amphioxus or lancelet) are derived directly from hemichordate enteropneusts. “Sensation interpretations” of fossils (Yunnunozoon, Cuthuymyrus) from Burgess Shale type deposits have added further confusion. Soft-part preservation of cono- dont animals, with V-shaped myomeres and a notochord, shows that they were segmented chordates, while probable eyes and teeth suggest that they were already on the vertebrate side.
Key words: Deuterostomes, protostomes, amphioxus, conodont animals, hemichordates, calcichordates, notoneuralians.
Zusammenfassung
Die Interpretation fruher Deuterostomia hinsichtlich ihrer Evolution und verwandtschaftlichen Beziehungen ist in den meisten Gruppen durch den Mangel an Weichkorpererhaltung sehr erschwert. Die kurzlich entdeckte Tatsache, daB Verte- braten, d. h. die einzigen echten Notoneuralia, im Gegensatz zu anderen bilateral symmetrischen Organismen eine mit ihrer ursprunglichen Oberseite nach unten gerichtete Lebensstellung einnehmen, hat zu der irrtumlichen Ansicht gefuhrt, daB alle Deuterostomia uber einen im Tierreich einzigartigen Bauplan verfugen. Diese Interpretation brachte naturgemaB jede Suche nach Verwandtschaftsverhaltnissen auf Abwege. Hinsichtlich der Echinodermata ist die bauplanmanige Variation in der Tat verwirrend. Die Interpretation einiger Fossilien mit Echinodermen-ahnlichen Kalzitskeletten als ,,calcichordate“ Vorfahren der Chordata setzt jedoch die hypothetische Rekonstruktion eines ungewohnlichen Bauplans sowie eine lange Serie hypothetischer Ubergange voraus. Die Anzahl der notwendigen Schritte ist sehr vie1 geringer, wenn Cephalochordaten (Amphioxus oder das Lanzettfischchen) von hemichorda- ten Enteropneusta abgeleitet werden. Zusatzliche Verwirrung hat es durch sensationelle Interpretationen von Fossilien, wie Yunnanozoon und Cathaymyrus aus Burgess-Schiefer-artigen Ablagerungen gegeben. Weichkorpererhaltung von Conodonten- tieren, die V-formige Myomere sowie einen Notochord besitzen, zeigen, daB es sich um segmentierte Chordata handelte, wahrend sie die Prasenz moglicher Augenstrukturen und Zahne bereits auf die Seite der Vertebraten stellt.
Schlusselworter: Deuterostomia, Protostomia, Amphioxus, Conodonten-Tiere, Hemichordata, Calcichordata, Notoneuralia.
Introduction latter are classically considered as echinoderms, but in a series of publications by Jefferies and Attempts to trace the invertebrate origin of ver- others they are placed on the vertebrate lineage. tebrates usually recognise the other chordate The history of research will not be considered groups, the cephalochordates and urochordates here, since it has recently been thoroughly re- (tunicates), as the closest relatives. Some Palaeo- viewed by Gee (1996). zoic skeletonised fossils, the calcichordates or The different phyla and subphyla of deuteros- carpoids, also play a role in the discussion. The tomes are notably different from one another in
Department of Palaeozoology, Swedish Museum of Natural History, PO. Box 50007, 104 05 Stockholm, Sweden. E-mail [email protected]> ’ Institut fur Zoologie, TalstraBe 33, D-04103 Leipzig, Germany. E-mail
Fig. 1. Downstrcarn and upstream collecting systems in marine larvae. In a typical trochophore larva, food particles collecl on the lee side of cilia. Cells are typically multiciliale, and cilia (long in drawing) compound. In a tornaria and echinoderm larva food particles are instead strained oll on the “windward” side of thc cilia. Cells are typically monociliate, and cilia (long in drawing) separate. In both cases the food particles are transported to the mouth by finc cilia (short in drawing) in the oral field. Water current solid black arrows, strained food particles stipplcd 84 Bergstrom, J., W. W. Naumann, J. Viehweg & M. Marti-Mus, Origin of Vertebrates by the larvae. Other alternatives include band demonstrates the difficulty or even inability to folding (several times among echinoderms, even elucidate relationships at the phylum level from with formation of “band islands”), as well as ad- the study of comparative anatomy and embryol- ditional rings similar to the telotroch (up to five ogy. The addition of phylogenetic (“cladistic”) rings in the doliolaria larva in sea cucumbers, see methods has not changed our inability to under- for instance Nielsen 1987: Fig. 31). Both systems stand the phylogenetic significance of phylo- could become enlarged in the equatorial plane, typic designs. This is not necessarily because the for instance by the downstream system in the ve- methods are poor, but because thcre wa5 an liger larva and the upstream system in adult overwhelming evolutionary parallelism between phoronids and pterobranchs. Additionally, the phyla. No formalised method enables us to upstream collecting system became improved by confidently sort out the order of innovations in the invention of pharyngeal slits in primeval deu- an evolutionary pattern dominated by parallel- terostomes, which allowed a more efficient se- ism. All approaches thus suffer from (a) the dif- paration of water and food particles before they ficulty to understand the significance of the in- enter the gut. termediate construction of tentaculate (or It is thus clear that there were several parallel oligomeran) phyla, and (b) the difficulty to ima- attempts to overcome the size limitation of the gine what evolutionary intermediates would downstream collecting system. Despite its simple have looked like. Nielsen’s approach docs not structure, or perhaps because of it, the upstream change this pattern. collecting system functions in larger larvae (for Salvini-Plawen (1982) was the first to inter- instance of Luidia) as well as in ciliary-feeding pret the morphologies of oligomerous, i.e., tenta- adults (bivalves, phoronids etc). Therefore pri- culate, animal phyla in terms of paedomorphosis. mary larvae with different adaptational ap- This was a radical but important step, since the proaches do not necessarily represent separate immense morphological and anatomical differ- monophyletic lineages but can be derived from a ences between sessile (oligomerous) and vagilc shared basic design. (polymerous and amerous) coelomates became Ciliary bands are underlain by nerve fibres. at once understandable as the result of reason- Garstang (1894) therefore suggested that the ably small genetic shifts. The new adult simply neural tube in chordates originated as a median utilised the ciliary feeding of the larva of its an- fusion between the paired dorsal ciliary bands as cestor. This made it unnecessary to assume an seen in enteropneust and echinoderm larvae. immense time interval for the transformation of, However, if vertebrates are dorso-ventrally in- say, a worm- or mollusc-like ancestor into an oli- verted compared with cnteropneusts, the larval gomerous tentaculated offspring. Furthermore, ciliary bands are on the side corresponding to Salvini-Plawen rightly saw the construction and the ventral side of vertebrates. development of phoronids as in several ways in- termediate between “spiralians” and deuteros- tomes. He also suggested that oligomerous ani- Deuterostome origins mals share a common ancestry, perhaps closest to the sipunculids, and that echiurids, annelids The differences between the protostomian and and arthropods are somewhat further apart. deuterostomian, or gastroneuralian and noto- Bergstrom (1986), being unaware of the ideas neuralian, conditions have been taken to indicate of Salvini-Plawen (1982), tried to understand the a more or less deep cleft between the two de- morphological and anatomical evolution ac- signs. In an extreme view, the two lines have cepting phylogenetic trees based on molecular been derived independently from coelenterates evolution. The apparent phylogenetic closeness (for instance, Nielsen 1987). The other extreme of animals with radically different body designs claims derivation of the deuterostomes from ad- asked for a non-conventional interpretation, and vanced protostomes. Other trees can be placed - just as in Salvini-Plawen’s approach - the nie- between these two extremes, deriving both deu- chanism was sought in paedomorphosis. ‘There is terostomes and “higher” protostomes from one difference, however. Whereas Salvini-Plawen either non-coelomate bilaterians, from coelo- saw the oligomerous/tentaculate condition as mates with few coelom compartments, or from being “invented” only once, the distribution of segmented polymeric coelomates. Some of these tentaculates in the molecular trees made Berg- alternatives have been illustrated by Salvini-Pla- strom believe in a multiple origination of tenta- wen (1982: Fig. 1). The multitude of alternatives culates. Mitt. Mus. Nat.kd. Berl., Geowiss. Reihe 1 (1998) ~ 85
When trying to identify the course of evolu- were generally derived from far-fetched compari- tion, we tend to look for unique events. Their sons with molluscs. Lamarck believed that the existence is suggested by character similarities ‘considerable gap’ between molluscs and verte- between animals and animal groups, and there brates would be filled by animals that ‘remained certainly have been such events. However, more to be discovered’. This gap was never filled ...”. often than not, phylogenetic trees change when Maybe it now is. Several molecular trees in fact characters used for the tree construction are indicate that molluscs are closer to deuteros- shifted. Obviously this must happen, since many tomes than is any other phylum (Fig. 2). Two characters are irregularly distributed among phy- centuries of anatomical comparisons had not im- la, and the distributional irregularity of one char- proved our understanding on this point. acter differs from that of virtually all the others. In fact, character distribution forms a mosaic in almost any tree. An attempt to use Nielsen’s “Calcichordates” (1987) character matrix in a students’ course for making a phylogenetic tree showed that at least If deuterostomes stem from soft-bodied ances- two thirds of the characters have multiple origin tors, and deuterostomes such as graptolites (in- and that each shift of trees made other characters cluding pterobranchs), enteropneusts and uro- polyphyletic. Nielsen’s matrix being highly gener- chordates lack a skeleton, it is easy to imagine alised, it is clear that most of the characters we that cephalochordates and vertebrates evolved use for tree-making are not homologous. Why, from ancestors devoid of a skeleton. However, then, should the paedomorphic origination of a an alternative derivation was suggested by T. Gis- tentaculate phylum have happened only once? 1Cn (1930), namely from ’carpoids’ (now called At least monophyly can not be stated as a fact. homalozoans) - an assemblage of extinct skele- Obviously, paedomorphosis at this level should tonised animals considered as echinoderms by always result in the development of ciliated ten- most echinoderm specialists. The idea was later tacles, sessility, simplification of the neural sys- adopted by R. P. S. Jefferies. In recent years he tem, a simple coelom with one compartment as- elaborated this hypothesis and also published a sociated with the tentacles, etc. The mere short review in French (Jefferies 1987). Recently presence of these characters does not prove a Gee (1996) gave another review of the hypoth- synapomorphy. Brachiopods, phoronids, bryozo- esis. ans, pterobranchs (graptolites), pogonophores Jefferies’ model starts with a pterobranch-like, and endoprocts are tentaculated phyla, but prob- exteriorly symmetrical, soft-bodied animal. From ably nobody regards their tentacles as a synapo- this, one line would have led to modern hemi- morphy shared by all groups. Endoprocts, for ex- chordates, another to the other deuterostomes. ample, are so tiny that they need neither a In the second line, animals reclined on one side coelom nor a vascular system. In this case, the and therefore turned asymmetric; they also de- tentacles appear to have evolved from the first veloped a calcitic mesodermal skeleton. Subse- ring of the larval cilia, whereas they apparently quently, one branch developed radial symmetry formed from the second ring in other tentacu- and turned into typical echinoderms. The other lates (but not in sipunculids, which are similar to branch developed a locomotive tail, thereby endoprocts in this respects but much larger, and turning into what Jefferies calls “calcichordates”. therefore cannot rely on ciliary feeding). Pogo- In the most agile forms, the body evolved into a nophores were once thought to be related to more symmetrical shape. The more asymmetric pterobranchs. However, chaetae and the segmen- forms belong to the order Cornuta, the - super- ted hind part of their body suggest that they are ficially - more symmetric ones to the order Mi- paedomorphic annelids. Also, brachiopods, phor- trata, both of the class Stylophora. According to onids, bryozoans and pterobranchs have enough Jefferies, such skeletonised forms have indepen- dissimilarities to suggest different origins - dently given rise to cephalochordates, urochor- although molecules indicate a close relationship dates and vertebrates. The stylophorans have a between brachiopods and phoronids. seemingly powerful appendage. Also, there was In conclusion, it is virtually impossible to re- in his interpretation a surprisingly large (mam- construct the origin of deuterostomes on the mal-size!) brain at the boundary between the base of classical comparative anatomy. A com- body and the appendage, from which coarse ment by Janvier (1996) may illustrate the dilem- nerve cords extended along the sides of the ma: “Early statements about vertebrate origins body. 86 Bergstrom, J., W. W. Naumann, J. Viehweg & M. Marti-Mus, Origin of Vertebrates
Annelida Origin of segmented chordates - Insecta three alternatives Crustacea Mollusca The urochordates (or “tunicates”) are so specia- Echinodermata lised that their present morphology is probably Vertebrata an unreliable guide to the organization of the A earliest chordates. For an analysis there remain the segmented chordates, that is, the cephalo- Annelida chordates (amphioxus) and vertebrates. These Insecta two groups, although dissimilar in many details, Mollusca share a sufficient number of characters including Vertebrata basic body design to reconstruct the shared an- B cestor. However, rather than hypothesising, we can also use the amphioxus anatomy for a direct Insecta comparison with hemichordate, and particularly Crustacea Mo Ilu sca enteropneust, anatomy. In the orientation of the Hemichordata animals given by nature, a cross section reveals a Echinodermata striking resemblance (Fig. 1; Bergstrom 1997: Vertebrata Fig. 3). Thus, gills open upwards, gonads are dor- C sal and placed in “wings” (not present in all en- teropneusts), the main nerve cord is ventral, the Nematoda stiffening notochord of amphioxus has a counter- Brachiopoda part (the pygochord) in the same position in Insecta some enteropneusts, the main body musculature Echinodermata is separated into paired ventral blocks, and the Vertebrata direction of blood circulation is identical. It may D be added that individual enteropneust species have been seen swimming and apparently feed- Insecta ing directly from the water. Crustacea At least three different origins for segmented Mollusca chordates have been suggested. According to Hemichordata one (Berrill 1955), they arose from sessile uro- Echinodermata chordates through paedomorphosis. According to Vertebrata Jefferies and his associates (see above), their ori- E gin is among so-called calcichordates. Ultimately, a derivation from enteropneust-like animals is Insecta supported by the similarities (Bergstrom 1997) Crustacea Mollusca seen when segmented chordates are oriented Echinodermata with the neural side down. Hemichordata Berrill’s and Jefferies’ models are both inge- Vertebrata nious in their own ways. The main drawback is F that they require many more evolutionary steps than a derivation directly from hemichordates Fig. 2. Molecular trees supporting a derivation of “deutero- and are therefore less parsimonious. This is an stomes” from protostomes. A, Phylogeny based on cytoch- rome c, mutational noise reduction method (redrawn from important difference, since in all three models, Bergstrom 1986). B, Phylogeny based on 5s rRNA, mutatio- and according to general views, the hemichor- nal noise reduction method (redrawn from BergstrOm 1986). dates are the least derived deuterostomes. The C, Phylogeny based on globin sequences (simplified from Goodman et a]. 1988). D, Phylogeny based on 18s rRNA lack of parsimony is particularly striking in Jeff- with secondary structure modcl of Dams et al. (from Winne- eries’ model (Table 1): the enteropneust design penninckx & Backeljau 1996). E, Phylogeny based on 18s has to be deformed and rotated sideways by 90°, rRNA with secondary structure model of Van de Peer et al. (from Winnepenninckx & Backeljau 1996). F, Strict consen- a new bilateral symmetry shaped 90” from the sus tree based on three 18s rRNA maximum parsimony old one, a hard skeleton has to be formed only trees published by Holland et al, with paired positions exclu- to be lost again, the tail has to be lost and ded (from Winnepenninckx & Backeljau 1996). Switching part of the tree upside-down may explain a striking differen- formed again, and so on until ultimately a body ce from certain other published trccs (for instance Raff 1995) design very close to that present at the start (in Mitt. Mus. Nat.kd. Berl.. Geowiss. Reihe l(1998) 87
Table 1. Comparison between the parsimonies in suggested consider the calcichordate appendage not as a transitions from hemichordates to segmented chordates, via calcichordates (Jefferies), via urochordates (Berrill), or di- tail, but as an anterior tentacle. Re-orienting the rectly (herein) body would be fatal to Jefferies’ interpretation since it is the very orientation that is fundamen- Jefferies Berrill herein tal to his idea. Yet, such a reorientation has re- loss of vagility in adult + cently received strong support by the recognition secondary vagility in adult + of two classes of skeletal elements in echino- adult moves backwards + adult secondarily moves + derms, the axials and the extraxials (that is, non- forwards again axials or abaxials; Mooi et al. 1994; David & loss of hemichordate basic + + - design Mooi 1996). The axials have distinct points of restoration of hemichordate + + - origin close to the mouth and are arranged in basic design rows, usually in pairs. The extraxials form irregu- formation of calcite skeleton + larly anywhere outside the axial fields. The axials reduction of calcite skeleton + reclining on side, with + also extend along the radii of the vascular sys- asymmetry tem and extend into arms, where such exist. In development of secondary + “calcichordates” the skeletal plates of the theca symmetry loss of one row of gill + obviously belong to the non-axial class, whereas openings at least part of the appendage is formed by ax- restoration of one row of + ials. Therefore the appendage corresponds in this gill openings rotation 90” of vascular system + respect to the circum-oral region and oral appen- rotation back 90” of vascular + dages in other echinoderms - not to the stalk. system removal of gills from mouth + end Cephalochordates return of gills to mouth end development of gut loop The amphioxus has many similarities with verte- reduction of gut loop loss of tail in adult brates. One of them is a tendency towards asym- development of new tail metry. It is important not to interpret this as a in adult similarity to asymmetric echinoderms including anus removed from tail + anus moved again to gastral + so-called calcichordates. What is shared between side chordates and echinoderms is the word ’asym- anus secondarily in tail metry’ in descriptions, not any factual asymmetry tail pulling tail pushing character. tail with dorso-ventral swing Amphioxus asymmetry is notable by the alter- tail with lateral swing nating dispositions of left and right muscle myo- functional exoskeleton return to functional endo- tomes and their nerves. It is also expressed in skeleton the development, during which gill openings first development of segmen- + + + form only on the left side (natural orientation, tation once segmentation independently + not conventional), and the mouth on the right 2nd time side, of the body (e.g., Bergstrom 1997: Fig. 4). somatocoel moved into + The significance of this phenomenon is un- tandem position somatocoel returned to pair t known. However, there may be some relation- position ship to the development of the notochord in the Number of steps 33 7 1 amphioxus, which extends to the anterior end to support it during burrowing. This extension, being present also in the embryo and larva, enteropneusts) is formed again. Jefferies’ model makes it impossible for the mouth to form in of chordate evolution obviously is the least likely front of the main ventral nerve cord, where it is of the three discussed here. This reminds us of located in urochordates, hemichordates and pro- critique to the same effect raised earlier, for in- tostomes. It is notable that the mouth of the stance by Jollie (1982). adult amphioxus is on the anterodorsal side of In order not to use Jefferies’ phylogenetic ar- the head, rather than on the ventral side as in guments on an interpretation of “calcichordates” urochordate larvae, hemichordates and many that he would not accept, his idea of their orien- protostomes. This deviation, however, may be an tation was followed in the comparison above. adaptation to filter-feeding, in which only the However, most specialists on fossil echinoderms mouth sticks out of the sediment. 88 Bergstrom, J., W. W. Naumann, J. Viehweg & M. Marti-Mus, Origin of Vertebrates
have the asymmetry. Although the mouth of Branchiostuma em- bryos forms in an unconventional position, it seems unlikely that it is not formed out of the old one. Anyway, it does develop in a new place (Fig. 4), and moves virtually to thc dorsal side in the adult. So it is anatomically justified to use the term ‘deuterostome’ for the amphioxus. This is in contrast to urochordates and hemichordates, where the mouth apparently develops in the same place as in yrotostomes. Apart from the am- phioxus, only the vertebrates are true ‘deuteros- tomes’ in the sense that they have their mouth / on the side that was originally the dorsal one Enteropneust Amp hioxus (Bergstrom 1997). However, while the am- phioxuses have retained their original dorsoven- Fig. 3. Idealized cross sections of an enteropneust heniichor- date and an amphioxus. Note the close correspondance be- tral orientation, so being functionally ‘gastro- tween the two when the amphioxus is not oriented as con- neuralians’, the vertebrates are turned over, to ventionally but, as in nature, with its “dorsal” side down become the only functional ‘notoneuralians’ among all animals. It is generally taken for granted that segmen- Perhaps the asymmetric mouth formation is tation in amphioxus and vertebrates originated caused by the development of the notochord. An as an adaptation to swimming. However, this is alternative possibility is that the ancestors of am- by no means certain. Amphioxus could probably phioxus animals did not burrow but lived as re- live without swimming in the adult stage and the cliners, with the mouth facing upward when the most important function of its segmented muscu- animals were lying on their left (natural) side. lature seems to be burrowing. Given the right As a third possibility, larval asymmetry may be a medium, this process is extremely fast, and it is functional requirement: before the tail is func- clear that this efficiency has to do with the tional, the larva of Branchiostoma lanceolatum powerful musculature. Burrowing lasts no more rotates while swimming, which means that it can than a few seconds at the most, and this appears collect food from all around its body when feed- to be the maximum time also for flashes of a fast ing. This behaviour is analogous to the rotational type of swimming. Actually, amphioxus swim- swimming in larvae of molluscs, annelids and en- ming may be considered as a kind of burrowing teropneusts. However, the larvae of Brunchiosto- performed in water. The repetition of neural ma floridae are reported not to rotate while nodes and muscle cells in appendiculariaceans feeding (Gilmore 1996; Stokes 1997) and still such as Oikopleura is sometimes compared with
3
1-3: successive mouth positions
Fig. 4. Generalised chordate cmbryo, with arrows showing the shift in mouth position from urochordates to Vertebrates. With the position acquired in vcrtebrates, the body has turned over so that the nerve cord became dorsal Mitt. Mus. Nat.kd. Berl., Geowiss. Reihe l(1998) 89
segmentation (polymery), but is probably better Here the conodont animal comes into the pic- referred to as pseudosegmentation (pseudo- ture. Its teeth cannot be homologised with verte- mery), that is, repetition of organs rather than of brate hard tissues (Schultze 1996). The same can, body packages, a logical pre-segmentation stage. of course, be said about modern cyclostome teeth and is thus no argument against a relation- ship with vertebrates. The soft parts of the cono- Conodonts and other early supposed dont animal are now better known and give a and real deuterostomes more conclusive answer to our questions. Purnell (1995) conveniently summarised the present Amphioxus has remained primitive in one im- knowledge of the soft parts. It is now clear that portant respect: its way of feeding. It still uses conodonts had a fish-like appearance, including a the ciliary feeding characteristic for both hemi- tail fin. The musculature was divided into V- chordates and urochordates. In vertebrates, such shaped myomeres, a character exclusively shared a mode of feeding is known only from larval by lancelets and vertebrates. The teeth worked lampreys, where it goes along with a semi-sessile laterally rather than vertically. This shows that life style. The distinguishing character of verte- the conodont animals were not gnathostomes. brates is that they search their food actively with Nevertheless, the mere presence of teeth as well the aid of sense organs situated in the head. as well-developed lateral eyes (or ears, as sug- Therefore eyes form an integral part of the basic gested by Pridmore et al. 1997) shows that the vertebrate design. Accordingly it should be pos- conodont animals were already on the vertebrate sible to use the presence or absence of eyes to side of the divide between vertebrates and lance- distinguish early lancelets from early non-skele- lets. There is no evidence of any hard skeleton, tonised members of the vertebrate lineage. but there was probably a thick notochord (Al-
A Conodont
B Pikaia
C Yunnanozoon
Fig. 5. A. Clydagrzathus windsorensis, a conodont animal from the Carboniferous of Scotland (from Bergstrom 1997). B. Pikaiu gracilens, a possible amphioxus relative from the Middle Cambrian Burgess Shale (from Bergstrom 1997). C. Yunnano- zoon lividurn, a worm-like animal from the Lower Cambrian Chengjiang fauna, supposed by some authors to be either a chordate or a hemichordate. Claimed muscle segments are here interpreted as hard, overlapping sclerites in the skin. A ventral area behind the mouth is supported by a flexible but strong dermal sheath. (New drawing. All three species about 4 cm long.) 90 Berrrstrom, J., W. W. Naumann, J. Vichweg & M. Marti-Mus, Origin ot Vertebrates
dridge & Purnell 1996, Fig. 3), whose lack of ver- recently been published (Chen et al. 1995; Dzik tebrae might cause a dispute whether or not con- 1995). Since repeated dark bands were inter- odont animals should be called vertebrates. How- preted as segmental myomeres, the animal was ever, this is of no interest or consequence for regarded as a chordate. However, in contrast to our understanding of conodont animals as being the segmental muscle blocks of segmentcd chor- early representatives of the vertebrate lineage dates these bands are not V-shaped (Fig. 5). which represents a sister group of the am- Furthermore, the bands overlap one another, phioxus lineage. There will also be a discussion which suggests that they are not muscle blocks, on whether or not the Early and Middle Cam- but hard sclerites in the skin. Also, the rows of brian conodont-like teeth really belonged to re- supposed internal gill slits (or branchial archs) latives of the later conodont animals. are not simply transverse to the gut, but shift The report of sensory organs for orientation from being more or less transverse to more or (eyes or perhaps ears) in conodont animals is less longitudinal, and from being straight to really important news. Much more dubious are strongly curved. This variability is extremely dif- other claimed chordates (Fig. 5). One of them is ficult to account for if the slits were connected the Burgess Shale (Middle Cambrian) Pikaia with dorso-ventral rows of penetrations of the gracilens. Its median fins and V-shaped myo- pharynx wall. Whatever Yunnanozoon may be, it meres make it a chordate, but there is no evi- has nothing to do with chordates (or hemichor- dence of either eyes or teeth. If this absence is dates: cf. Hou et al. 1991, pp. 408-409). real, Pikaia may be a relative of the modern am- Also from the Chengjiang fauna Shu et al. phioxus. Its size and shape possibly indicates that (1996) described a single, poorly preserved, spe- it was not a burrower, but may have been swim- cimen of Cathaymyrus diadexus and interpreted ming or reclining on its side on the bottom. Any- it as a cephalochordate. This is possible, but since way, the clayey sediment in which it is preserved no fins are preserved and the interpretation of could not have been suitable for burrowing. the alimentary canal, notochord, pharynx, gill Reinterpretations of Yunnanozoon from the slits, and myomeres must be corroborated by Lower Cambrian Chengjiang Lagerstatte have more and better material, it is not yet possible to make any reasonable judgement on the sys- tematic position or importance of this species. Echinodermata
Graptolithina (incl. modern pterobranchs) Deuterastome phylogeny Enteropneusta 4- As a result of the considerations, the phylogeny Urochordata of the deuterostomes can be summarised as in Fig. 6. It is comparatively easy to imagine how 8- Cephalochordata both echinoderms and enteropneusts could have evolved from pterobranch-like animals. Chor- Vertebrata 9- dates are most reasonably derived from enterop- Fig. 6. Suggested cladogram of deuterostomian phyla. 1, ori- neust-like ancestors (cf. Fig. 3). Since in effect gination of the Graptolithina (and Deuterostomia as a whole): paedomorphic retension of ciliary feeding and deve- only cephalochordates and vertebrates have a lopment of tentacles, development of pharyngeal slits as wa- secondary position of the mouth, and only the ter outlets; three coelomic compartments, and hydropore, de- vertebrates are turned upside-down, the other velopment of tube, and U-shaped gut; 2, detachment from groups may as well be called protostomcs or gas- tube, reclining with initial asymmetry, calcite endoskeleton; 3, detachment from tube, burrowing with the use of proboscis, troneuralians. This means that there is every rea- and straightening of gut; 4, no major change; 5, swimming son to believe that they, like other ciliary-feed- with development of notochord (from enteropneust pygo- ers, were derived from chord?) and tail, loss of proboscis and hydropore, develop- other protostomes by mental modification of blastopore closure resulting in tube- paedomorphosis. Vertebrates are thc only real shaped neural cord; 6, simplification, loss of constant directi- notoneuralians. on of blood flow, loss of coelom, strong development of cuti- cle and loss of external ciliation; 7, development of powerful segmented muscles for new burrowing habits, segmentation affecting 3rd coelom pair, shift of mouth to dorsal side; 8, Acknowledgements improvement of adaptation; 9, loss of ciliary feeding, deve- lopment of sensory organs including eyes with adoption of We are grateful Professor Ragnar Olsson (or inspiring discus- active search for food, rotation to upside-down posture to sions, to the Marine Biological Station at Kristincberg for keep mouth below eyes. logistics, and to the Hierta-Retzius Foundation and the foun- Mitt. Mus. Nat.kd. Berl., Geowiss. Reihe 1(1998) 91 dation “Till Broderna Jacob och Marcus Wallenbergs Minne” vertebrates involving sog and chordin. - Nature 376 for economic support. We are also grateful to the reviewer 249-253. Professor A. Seilacher for suggesting numerous improve- Hou Xian-guang, Ramskold, L. & Bergstrom, J. 1991. Com- ments of our manuscript. position and preservation of the Chengjiang fauna - a Lower Cambrian soft-bodied biota. - Zoologica Scripta 20(4): 395-411. Janvier, P. 1996. Spineless debates. - Nature 384 324-325. Referencs Jefferies, R. P. S. 1987. A l’origine des vertkbres: les animaux bottines. - La Recherche 18: 1101-1104. Aldridge, R. A. & Purnell, M. A. 1996. The conodont contro- Jollie, M. 1982. What are the ‘Calcichordata’? and the larger versies. - Tree 11 (11):463-468. question of the origin of chordates. - Zoological Journal Arendt, D. & Niibler-Jung, K. 1994. Inversion of dorsoven- of the Linnean Society 75: 167-188. tral axis? - Nature 371: 26. Lacalli, T. C. 1996. Landmarks and subdomains in the larval Chen Jun-yuan, Dzik, J., Edgecombe, G. D., Ramskold, L. & brain of Branchiostomu. Vertebrate homologs and inver- Zhou Gui-qing 1995. A possible Early Cambrian chor- tebrate antecedants. - Israel Journal of Zoology 42, Sup- date. - Nature 377 720-722. plement, The Lancelets: A new look at some old beasts: Bergstrom, J. 1986. Metazoan evolution - a new model. - S131-S146. Zoologica Scripta 15(3): 189-200. Mooi, R., David, B. & Marchand, D. 1994. Echinoderm ske- - 1994. Ideas on early animal evolution. In Bengtson, S. letal homologies: Classical morphology meets modern Early Life on Earth, pp. 460-466. Columbia University phylogenetics. In David, B., Guille, A., FCral, N. & Roux, Press, New York. M. (eds). Echinoderms through Time: 87-95, Balkema, - 1997. Origin of high-rank groups. - Palaeontological Re- Rotterdam. search 1: 1-14. Nielsen, C. 1987. Structure and function of metazoan ciliary Berrill, N. J. 1955. The Origin of Vertebrates. 237 pp. Claren- bands and their phylogenetic significance. - Acta Zoolo- don Press, Oxford. gica 68: 205-262. David, B. & Mooi, R. 1996. Embryology supports a new Niibler-Jung, K. & Arendt, D. 1994. Is ventral in insects dor- theory of skeletal homologies for the phylum Echinoder- sal in vertebrates? - Roux’s Archives Developmental mata. - Comptes Rendus de 1’Academie des Sciences de Biology 203: 357-366. Paris, Sciences de la vie 319: 577-584. Pridmore, P. A., Barwick, R. E. & Nicoll, R. S. 1997. Soft De Robertis, E. M. & Sasai, Y. 1996. A common plan for anatomy and the affinities of conodonts. - Lethaia 29: dorsoventral patterning in Bilateria. - Nature 380: 37-40. 317-328. Dzik, J. 199.5. Yunnanozoon and the ancestry of chordates. - Purnell, M. A. 1995. Large eyes and vision in conodonts. - Acta Palaeontologica Polonica 40: 341-360. Lethaia 28: 187-188. Emlet, R. B. 1994. Body form and patterns of ciliation in Raff, R. A. 1994. Developmental mechanisms in the evoluti- nonfeeding larvae of echinoderms: Functional solutions to on of animal form: Origins and evolvability of body plans. swimming in the plankton? - American Zoologist 34 In Bengtson, S. Early Life on Earth, pp. 489-500. Colum- 570-585. bia University Press, New York. Garstang, W. 1894: Preliminary note on a new theory of the Salvini-Plawen, L. 1982. A paedomorphic origin of the oligo- phylogeny of the Chordata. - Zoologische Anzeiger 17: merous animals? - Zoologica Scripta ll(2): 77-81. 122-125. Schultze, H.2 1996. Conodont histology: an indicator of ver- Gee, H. 1996. Before the Backbone: Views on the Origin of tebrate relationship? - Modern Geology 20: 275-285. the Vertebrates. 346 pp. Chapman & Hall. Shu De-gan, Conway Morris, S. & Zhang Xing-liang. 1996. A Gilmore, T. H. J. 1996. Feeding methods of cephalochordate Pikaia-like chordate from the Lower Cambrian of China. larvae. - Israel Journal of Zoology 42, Supplement: - Nature 384: 157-158. S87-S95. St. Hilaire, E. G. 1822. ConsidCrations gCnCrales sur la ver- GislCn, T. 1930. Affinities between the Echinodermata, Ente- tkbres. - MCmoires de MusCum Histoire Naturelle 9: ropneusta, and Chordonia. - Zoologiska Bidrag fr5n 89-119. Uppsala 12: 199-304. Stokes, M. D. 1997. Larval locomotion of the lancelet Brun- Goodman, M., Pedwaydon, J., Czelusniak, J., Suzuki, T., Go- chiostoma floridae. - The Journal of Experimental Biolo- toh, T., Moens, T., Shishikura, F., Walz, D. & Vinogradov, gy 200: 1661-1680. S. 1988. An evolutionary tree for invertebrate globin se- Winnepenninckx, B. & Backeljau, T. 1996. 18s rRNA align- quences. - Journal of Molecular Evolution 27: 236-249. ments derived from different secondary structure models Holley, S. A., Jackson, P. D., Sasai, Y., Lu, B., De Robertis, can produce alternative phylogenies. - Journal of Zoolo- E. M., Hoffmann, F. M. & Ferguson, E. L. 1995. A con- gical Systematics and Evolutionary Research 34 served system for dorsal-ventral patterning in insects and 135-143.