SUPPLEMENTARY INFORMATION

Transcription

1 doi: /nature12980 Terminology: nasal capsule vs. rostral capsule Different anatomical terms have been used in the past to denote the distinct fissure-bounded nasal region of placoderms. Here we apply the following terminology. The rostral capsule corresponds to the dermal bony part of the capsule (i.e. the rostral and pineal or rostropineal plates); the nasal capsule corresponds to the underlying perichondrally ossified structure wrapping the lamina cribrosa. The whole assemblage is often called either nasal or rostral capsule in the literature. Here, for precision, we prefer the term rostronasal capsule. Supplementary specimens The rostronasal capsule of Romundina stellina could be virtually reconstructed on the basis of two specimens: the isolated capsule GSC figured by Ørvig , and MNHN CPW 13 (Extended Data Figure 4), a skull with an attached rostronasal capsule more recently discovered by one of us (D.G.). The shape of the dermal bone of the capsule differs between the two specimens (jigsaw-piece like in GSC 11515; more trapezoidal in CPW 13). However, no information on the internal view is available for CPW 13 as this specimen has not yet been scanned. As a consequence, we decided to recreate the external shape of the rostronasal capsule (i.e. dermal bone) according to CPW 13, slightly modified to fit with the model of CPW 1. The ventral and posterior sides (and so the anterior most part of the telencephalon, including the terminal nerve) have been recreated according Ørvig s figures 13. Supplementary methods: Acquisition and reconstruction of the PPC-SRµCT data set The scan data set has an isotropic voxel size of 7.46 µm and was made with a monochromatic beam with an energy of 51 kev using a double Si111 crystal monochromator in Bragg reflection. The distance between sample and detector was 900 mm projections over 360 degrees were taken with a FreLoN CCD camera, with 0.9 seconds of exposure time per projection. In order to image the complete specimen, four scans covering 7 mm vertically each were performed. The slices were reconstructed using filtered back projection algorithm (PyHST software, ESRF). The reconstructed slices were then converted into 16 bits tif image stacks that were concatenated to obtain a single stack covering the whole sample. In order to reduce the data size for general anatomical observations, a second version of the reconstructed scan was calculated with a 2*2*2 binning and an 8 bits conversion. Supplementary methods: Anatomical reconstruction In addition to structures (bone and space fills) that have been modelled as preserved, the final 3D model includes areas of reconstruction produced by left/right mirroring of structures that are only preserved on one side, and some prosthetics. Mimics v.12.3 and v.13.1 (Materialise) were used for the 3D modelling (segmentation, 3D object rendering and STL polygon creation). Maya Autodesk 2012 was used for extra editing of the STLs (color, texture, corrections) and kinematics, as well as for mirroring and the construction of prosthetics. 1

2 Mirroring, which has been particularly useful as a remedy for information loss caused by the oblique break across the specimen, requires a certain amount of positional and morphological adjustment because the specimen is not perfectly bilaterally symmetrical. The prosthetics of the rostronasal capsule and other reconstructed structures (for example the nasal sacs and the anterior face of the telencephalon) have all been given smooth, geometric surfaces to distinguish them from actual scanned anatomy. All are based on preserved correlates, such as cavities in the rostropineal capsule or nerve grooves on the subocular shelves of the braincase. Supplementary discussion: inferring non-observable characteristics in fossil taxa. The arguments presented in this paper rest substantially on inferences about certain attributes of fossil vertebrates that cannot be observed directly. They fall into two groups: firstly, the relative size, shape and position of different regions of the brain and associated structures; and secondly, the size, position and migratory pathways of certain populations of ectomesenchyme, i.e. mesenchymal cell masses of neural crest origin. Because these inferences are central to the paper we present here a detailed account of the sequence of logical steps by which they have been made. General considerations: 1) Certain attributes are present both in extant cyclostomes and gnathostomes, and are uncontroversially homologous between these two clades, indicating that they are general for crown-group vertebrates as a whole. These general vertebrate attributes include: an infraoptic ectomesenchyme subdivided into premandibular and mandibular components; a supraoptic ectomesenchyme; absence of mesodermal chondrification anterior to the notochord; nasal sacs with olfactory epithelia; a hypophysis; and a brain comprising inter alia a telencephalon carrying the terminal (0) and olfactory (I) cranial nerves and a diencephalon carrying the optic (II) nerve and hypophysis 1-8,29. 2) The fossils discussed in this paper belong to the gnathostome stem group and thus fall within the phylogenetic bracket formed by extant cyclostomes and gnathostomes 9-12,20. Furthermore, their phylogenetic placement rests in part on possession of skeletal characters which are proxies for some of the attributes mentioned in 1), such as a cranial cavity with lateral openings corresponding to the standard set of vertebrate cranial nerves, associated with nasal cavities and a hypophysial recess. We thus infer that the fossils possessed these general vertebrate attributes. The brain: 3) Following from 1) and 2) we assume that, in the brains of the fossil taxa as in extant vertebrates, the optic (II) nerve marked the anterior margin of the diencephalon and the posterior edge of the hypothalamus marked the posterior margin. The cranial cavities of the fossils carry numerous nerve canals and outpocketings that can be uncontroversially identified with those of extant vertebrates, including an optic nerve foramen and recesses for the hypothalamus and hypophysis. Anterior to the optic nerve foramen, the telencephalic cavity is extremely short. Furthermore, the hypophysial recess is 2

3 directed anteroventrally (Shuyu 11, Brindabellaspis 18, Romundina) or ventrally (Dicksonosteus 15,22 ), and the distance between the hypophysial recess and nasal sacs is much smaller than in extant gnathostomes. These shape characteristics match the brains of extant cyclostomes 29 (Figure 3 of paper). We infer that the brains of the fossils had proportions matching those of their cranial cavities, i.e. the telecephalon was short and the hypophysis directed anteroventrally or ventrally. This inference is not entirely unproblematic, because both extant (Latimeria, a sarcopterygian) and fossil (Sibirhynchus, a stem holocephalan) examples show that the adult brain can be smaller than the cranial cavity and different in shape However, in Sibirhynchus the optic (II) nerve and hypophysis line up with the corresponding opening and recess in the cranial cavity wall 33, and comparison with extant holocephalans suggests that a very long unpreserved telencephalon would have connected the preserved diencephalon with the nasal sacs 34. The longitudinal proportions of the brain are thus similar to those of the cranial cavity, even though the brain is slender and does not fill the available space. In Latimeria, the brain fills the cranial cavity in early juveniles, but in adults it is extremely small and confined to the posterior part of the cavity because it fails to grow at the same rate as the skull 31,32. This gives rise to morphological peculiarities in the adult brain, such as extremely long olfactory (I) tracts, anteriorly directed optic (II) nerves, and an anteriorly directed hypophysis. However, with these exceptions the basic proportions of the brain match those of the cranial cavity, showing that the strange adult condition is simply a very drastic allometric effect. In contrast to this, accommodating a brain with a long gnathostome-like telencephalon in the cranial cavity of one of our fossils would require the brain to be not only much smaller than the cavity but also quite differently proportioned. There is no precedent for this in extant vertebrates, and it would leave the shape of the cranial cavity unexplained unless the brain had undergone dramatic shape change during growth. We regard this interpretation as implausible and conclude that the cranial cavities of our fossil stem gnathostomes reflect the presence in these animals of brains with short telencephala and anteroventrally to ventrally directed hypophyses. The neural crest composition of the face: 4) On basis of points 1) and 2) we infer that the fossil taxa had facial skeletons composed of some combination of premandibular infraoptic ectomesenchyme, mandibular infraoptic ectomesenchyme and supraoptic ectomesenchyme, without a mesodermal contribution. The specific composition of each facial skeleton cannot be inferred from phylogenetic bracketing, because extant cyclostomes and gnathostomes have contrasting morphologies that reflect different cell population behaviours, and our fossils fall between these extant groups rather than within either of them. Reasoning from simple morphologicaldevelopmental parsimony, we assume that the occurrence in the fossils of either cyclostomelike or gnathostome-like morphological components respectively reflect the presence of corresponding cyclostome-like or gnathostome-like patterns of ectomesenchymal growth. Thus, the presence of a nasohypophysial duct indicates that premandibular ectomesenchyme grew forwards as a pair of lateral posthypophysial processes like in cyclostomes, whereas a solid trabecular region indicates that the posthypophysial processes met in the midline anterior 3

4 to the hypophysis like in a gnathostome. Similarly, a gnathostome-like nasal region with laminae cribrosae and an internasal septum is assumed to be formed from supraoptic ectomesenchyme like the corresponding structures in an extant gnathostome. Palatoquadrates are assumed to be composed of mandibular infraoptic ectomesenchyme. These assumptions lead to the following interpretations of the fossil taxa. Shuyu: As a dorsally positioned nasohypophysial duct is present, flanked by a cyclostome-like upper lip without associated palatoquadrates, the posthypophysial processes must have grown forward laterally around the nasal and hypophysial placodes. This also implies that the facial skeleton lateral and anterior to the duct is composed principally of infraoptic ectomesenchyme (contra Oisi et al. 7 ). However, the presence of a small median trabecular process between the hypophysial recess and olfactory bulbs 11 indicates that a minor subset of the premandibular ectomesenchyme followed a gnathostome-like trajectory, separating from the posthypophysial processes and converging in the midline anterodorsal to the hypophysis. Supraoptic ectomesenchyme was probably restricted to the posterior margin of the nasohypohysial duct, like in extant cyclostomes, and mandibular ectomesenchyme probably did not grow forwards lateral to the posthypophysial process. However, the absence of skeletal boundaries 11 makes it impossible to determine the exact extent of these two cell populations. Romundina and Kujdanowiaspis: These two taxa have identically constructed skulls, differing only in the proportions of certain elements 15,21,22. The foramen for nerve VII(hm), at the tip of the anterior postorbital process just behind the orbit, is flanked by a palatoquadrate articulation anteriorly and two hyoid arch articulations (hyomandibular and opercular cartilages) posteriorly. This shows that it marks the level of the first pharyngeal pouch, which is the posterior boundary of the trigeminal neural crest stream in extant vertebrates. Anterior to this foramen and the anterior tip of the notochord, landmarks that form an essentially straight transverse line across the head, the entire skull must be formed from trigeminal ectomesenchyme, i.e. from a combination of supraoptic, premandibular infraoptic, and mandibular infraoptic ectomesenchyme. The presence of a solid trabecular region without a nasohypophysial duct indicates that premandibular ectomesenchyme followed a gnathostomelike growth trajectory towards the midline in front of the hypophysis, below the telencephalon and nasal sacs. Mandibular ectomesenchyme grew forward alongside the premandibular ectomesenchyme, as shown by the presence of palatoquadrates articulating with the trabecular region. The rostronasal capsule contains laminae cribrosae and an internasal septum, indicating that it is composed of supraoptic ectomesenchyme. Because the rostronasal capsule is sharply demarcated from the trabecular region, and separated from it by an optic fissure, we infer that the entire trabecular region including the projecting precerebral part and the suborbital shelves, is composed entirely of infraoptic ectomesenchyme with no supraoptic contribution. The optic fissure anterior to the nerve II foramen is interpreted as the line of contact between supraoptic and premandibular infraoptic ectomesenchyme. Kujdanowiaspis (and Dicksonosteus 22 ) differs from this organization only in that the trabecular-ethmoid region is much shorter anteriorly so that there is no prenasal upper lip. 4

5 Supplementary discussion: the maxillary branch of the trigeminal nerve The maxillary branch of the trigeminal nerve has been labelled as V3(mx) in our illustrations (Extended Data Figure 3), that is as belonging to the third branch of he trigeminal nerve. However, it is generally accepted that the third branch of the trigeminal nerve carries the mandibular ramus, whereas the second branch carries the maxillary ramus. The reason for this unorthodox labelling is that a groove for the V3(mn) ramus is visible on the orbital shelf, and is connected to a smaller anteriorly directed groove that must have housed the maxillary ramus. However, because the evidence consists of grooves and not the nerves themselves, we cannot rule out the possibility that the maxillary ramus in fact connected to the second branch of the trigeminal nerve; unfortunately, the limited evidence for the trajectory of V2 after leaving the trigeminal recess does not allow us to confirm or refute this hypothesis. Phylogenetic Analysis 'Posterior-nose' forms like Romundina, Brindabellaspis and antiarchs have been interpreted as phylogenetically basal placoderms in all analyses published over the last few decades, irrespective of whether the Placodermi as a whole have been viewed as monophyletic 20 or paraphyletic 9,10,12. However, the most recent analyses 9,10,12 have not included Romundina. In order to remedy this and test the assumption that Romundina is a relatively primitive placoderm, we have performed a phylogenetic analysis using a modified version of the recent data matrix by Zhu et al incorporating Romundina (see supplementary data: Dupret_SI_data_matrix_NDE, or Dupret_SI_data_matrix_McClade) and the basal phyllolepid Gavinaspis 35. The total number of taxa is 77, scored for 255 characters. Two character codings for Galeaspida have been changed in phylogenetically significant ways (in addition to a number of changes from "?" to "-" which are descriptively appropriate but have no impact on the analysis): character 86 (sinus superior) has been recoded from "0" (absent) to "1" (present) on basis of data from Gai et al. 11 ; and character 115 (endoskeletal postbranchial lamina on scapular process) has been recoded from "0" (present) to "-" because galeaspids have no scapulocoracoid at all. Two new characters have been added (at the end of the list for convenience): #254 to the presence or absence of an articulated jaw, and #255 to the presence or absence of an endocranial optic fissure.character #169 has been significantly modified from two states of characters to five (in order to illustrate the diversity of dermal craniothoracic articulation among placoderms). Description of the retained 50% majority rule consensus tree (see Extended Data Figure 1): A heuristic research (simple stepwise addition, TBR swapping algorithm, maximum number of trees = 30000, 1000 replicates) leads to most parsimonious trees of 614 steps each (CI=0.4332; RI=0.8119). Osteostraci is sister group for Galeaspida + Gnathostomata. The relationships at the base of the strict consensus tree (Extended Data Figure2) are the same as those described in Zhu et al. 12 (polytomy between Parayunnanolepis, Euantiarchi, Brindabellaspis, Macropetalichthys, rest of the tree). In the 50% majority rule consensus tree (Extended Data Figure 1), Antiarchi are monophyletic and the most basal jawed vertebrate clade. Macropetalichthys is sister group of Brindabellaspis and this clade is sister group with 5

6 the rest of the Gnathostomata, i.e. a clade composed by (Romundina (Arthrodira (Entelognathus (Ptyctodontida, Eugnathostomata)))). EUgnathostomata is sensu ref. 36. The relationships more coronal than the Placodermi are slightly different than those described in Zhu et al. 12, because Acanthhodians appear as the monophyletic sistergroup of Chondrichthyes, rather than a paraphyletic stem group of the latter. The intra-chondrichthyan relationships are slightly different too (a polytomy appears between (Doliodus, Debeerius, (Cladodoides, Orthacanthus)) while these relationships are resolved in Zhu et al. 12 (supplementary figure 2a,b). The relationships among Osteichthyes are a bit more puzzling, since the topology of the present analysis resembles more that of Zhu et al. 12 without Entelognathus (Zhu et al. 12, supplementary figure 2b). However, these taxa are beyond the scope of this article and their interrelationships will not be discussed here. In the following description of character state changes in the 50% majority rule tree, superscript stands for change to derived state; subscript stands for reversion; bold stands for non-ambiguous transformation. The primitive state is determined by outgroup comparison; in case of multistate characters, the immediate ancestral state is specified in the text. The polytomy at the base of the Gnathostome tree in the strict consensus of the present analysis (when opposed to previous studies by Davis et al. 10 ) can be explained by different reasons: Parayunnanolepis possesses pelvic fins (121 1 ) contrary to the outgroup taxa (Osteostracans and Galeaspids) and the Euantiarchi 37 ; Semidentine is considered as a gnathostome symplesiomorphy (5 1 ; but dentinous tissue is lost in Euantiarchi 4 0 ); this condition is unknown in Parayunnanolepis. Despite the presence of a dermal armour for the pectoral fins and the presence of a lateral plate, and because of the two previously discussed characters, the Antiarchi are put in polytomy with other basal placoderms (Brindabellaspis and Macropetalichthys) rather than a clade. The discussed 50% majority rule consensus nevertheless shows that monophyletic Antiarchi appear in 63% of the trees. Many important characters are coded as non applicable in the outgroup: o Dermal bones of the skull roof (when ostracoderms have a one component shield; character 23), consolidated cheek plates and sensory line system pattern (character 24, ); o Branchiostegal series- and opercule-related characters (characters 28-32); o Gular plate related characters (36) o Hyal elements (37-38) o Jaw and teeth related characters (39-53, 80-81, 95; , ) o Spiracular groove (62) o Metotic fissure (71) o Blood vessels aspects (77-79) o External semicircular canal (87) o Dermal thoracic armour (108) o Scapulocoracoid ( , ; ) o Fins, fin spines, claspers etc. (120, 122, 129, , ) 6

7 o Typical tissues of other clades/groups absent ( ) o Scales ( ) o Dermal craniothoracic articulation (169) o Taking those? and - into account, the optimizable amount of data is of 68% for Osteostraci (140 - and 34? ) and 76% for Galeaspida (160 - and 35? ). Node 1 to node 2 (Gnathostomata) 5 1 (Dentine kind; semidentine; CI = 0.400) 7 1 (Lepidotrichia or lepidotrichia-like scale alignment absent; CI= 0.200) 18 0 (Dermal skull roof includes large dermal plates; CI = 0.500) 19 1 (Tesserae morphology: (microsquamose, not larger than body tesserae; CI = 0.200) 33 1 (Opercular (submarginal) ossification present; CI = 0.500) 85 1 (External (horizontal) semi-circular canal present; CI = 1) (Pelvic fins present; CI =.500) (jaws present; CI = 1) (optic fissure present; CI = 0.5) Node 2 to Node 3 (Antiarchi; found in 63% of the trees) (Pectoral fins covered in macromeric dermal armour present; CI = 1) (Lateral plate present; CI = 1) Node 3 to node 4 (Euantiarchi) 4 0 (Dentine absent; CI = 0.500) (Pelvic fins absent; CI = 0.500) (Pectoral fin base has large, hemispherical dermal component present; CI = 1) Node 2 to node 5 (unnamed clade); 75% of the trees 25 1 (Pineal opening perforation in dermal skull roof absent; CI = 0.250) 39 1 (Oral dermal tubercles borne on jaw cartilages present; CI = 0.167) (Synarcual present; CI = 0.500) (Two pairs of paranuchal plates; CI = 0.333) (Type of dermal neck-joint spoon-like, ptyctodont and petalichthyid condition in Goujet and Young ; CI = ancestral state was 2, reversed ginglimoid (antiarch condition) (common ventral fenestra for anterior and posterior nostrils; CI = 0.500) It is noteworthy that two pairs of paranuchal plates are observed in very basal arthrodires (i.e. Yujiangolepis 38 ), petalichthyids (e.g. Lunaspis 39, Eurycaraspis 40 ) but also early ptyctodontiforms, like Hagiangella

8 Node 5 to node 6 (unnamed clade; Macropetalichthys + Brindabellaspis; 50% of the trees) 16 1 (Sensory line network passes through canals enclosed within dermal bones; CI = 0.250) 24 0 (Consolidated cheek plate absent; CI = 0.250) 77 1 (Canal for lateral dorsal aorta within basicranial cartilage present; CI = 0.250) (Contact of nuchal or centronuchal plate with paired preorbital plates present; CI = 0.500) Node 6 to Brindabellaspis (Type of dermal neck-joint as simple contact; CI = 0.800) Node 6 to Macropetalichthys 56 1 (Nasal opening(s) ventral and anterior to orbits; CI = 0.500) 57 1 (Olfactory tracts elongate and tubular (much longer than wide; CI = 0.250) 79 1 (Canal for efferent pseudobranchial artery within basicranial cartilage present; CI = 0.500) (Large unpaired median median skull roofing bone anterior to the level of nasal capsules absent; CI = 0.500) The character complement of Macropetalichthys includes some characters shared with much more derived gnathostomes: long olfactory tracts and nasal openings ventral and anterior to orbits. This may induce its unresolved position relative to Brindabellaspis and Antiarchi. It is sister group of Brindabellaspis in 50% of the trees. Node_5 to 7 (unnamed clade) 11 1 (Body scale profile flattened; CI = 0.200) 55 1 (Median dermal bone of palate (parasphenoid) present; CI = 0.333) 66 1 (Orbit dorsal or facing dorsolaterally, surrounded laterally by endocranium; CI = 0.500) 76 1 (lateral wall of jugular vein canal incomplete or absent; CI = 0.500) 86 1 (Sinus superior present; CI = ) 95 1 (Position of hyomandibula articulation on neurocranium posterior to orbit; CI = 0.500) (Ventral notch between parachordals present or entirely unfused; CI = 0.250) (Posterior nostril not associated with orbit; CI = 0.500) (Posterior end of supraorbital canal in parietal ancestral state 3 in nuchal plate; CI = 0.444) Node 7 to Romundina 25 0 (Pineal opening perforation present; CI = 0.250) 8

9 80 1 (Position of basal/basipterygoid articulation: anterior to hypophysial opening; CI = 0.500) 87 1 (External (horizontal) semicircular canal joins level with posterior ampulla; CI = 0.500) (Pectoral fenestra completely encircled by dermal shoulder armour absent; CI = 0.333) (Posterior end of supraorbital canal in postpineal plate; CI = 0.444) (Anterior portion of preopercular canal absent; CI = 0.500) (Fenestra ventrolateralis absent; CI = 0.500) (Dorsal cleithrum (AL of the Placodermi), ventral cleithrum (AVL of the Placodermi) and pectoral spine (SP of the Placodermi); CI = 0.333) (Endoskeletal supports in pectoral fin is a single element ("humerus") articulating with girdle; CI = 0.500) Node 7 to node 8 (unnamed clade) 23 1 (Series of paired median skull roofing bones that meet at the dorsal midline of the skull (rectilinear skull roof pattern present; CI = 0.333) 47 1 (Insertion area for jaw adductor muscles on palatoquadratelateral; CI = 0.500) 56 1 (Nasal opening(s) ventral and anterior to orbits; CI = 0.500) 58 1 (Prominent pre-orbital rostral expansion of the neurocranium absent; CI = 1) 69 1 (Main trunk of facial nerve (N. VII) is stout and divides within otic capsule at the level of the postorbital process; CI = 1) (Large unpaired median skull roofing bone anterior to the level of nasal capsules : absent; CI = 0.500) (Dermal neck-joint is sliding type, with dermal shoulder girdle plate with flat articular flange, actinolepid-type ; CI = 0.800) (Most posterior major bone of cheek bearing preopercular canal (preopercular) extending forward, close to orbit present; CI = 0.500) (Developed postorbital cavity present; CI = 0.500) (Pectoral propterygium present; CI = 0.333) Node 8 to node 9 (Arthrodira): 17 1 (Jugal portion of infraorbital canal joins supramaxillary canal absent; CI = 0.333) 22 1 (Endolymphatic ducts with oblique course through dermal skull bones present; CI = 1) 64 1 (Subpituitary fenestra present; CI = 0.500) 86 0 (Sinus superior absent or indistinguishable from union of anterior and posterior canals with saccular chamber; CI = 0.333) (Scapular process of shoulder endoskeleton absent; CI = 0.500) (Fin base articulation on scapulocoracoid eurybasal; CI = 0.250) 9

10 162 0 (One pair of paranuchal plates; CI = 0.333) (Ethmoid articulation for palatoquadrate extends posteriorly to the level of N.II., CI = 0.333) Node 9 to node 10 (Phyllolepida + Phlyctaenioidei) 11 0 (Body scale profile shows distinct crown and base demarcated by a constriction (neck); CI = 0.200) 34 1 (Shape of opercular (submarginal) ossification: narrow, rod-shaped; CI = 0.333) (Posterior process of the paranuchal plate behind the nuchal plate (dorsal face) present; CI = 0.500) (Junction of posterior pitline and main lateral line close to posterior margin of skull roof; CI = 0.500) (Pelvic girdle without substantial dermal component; CI = 0.333) Node 10 to node 11 (Phyllolepida) 23 0 (Series of paired median skull roofing bones that meet at the dorsal midline of the skull (rectilinear skull roof pattern absent; CI = 0.333) (Pectoral fenestra completely encircled by dermal shoulder armour absent; CI = 0.333) (Contact of nuchal or centronuchal plate with paired preorbital plates present; CI = 0.500) (Position of middle and posterior pit lines close to midline; CI = 0.200) In this analysis, the Phyllolepida are sister group of the Phlyctaenioidei, and together with Sigaspis illustrate a paraphyletic condition of the Actinolepidoidei similar to that found by Dupret and Dupret et al , in contrast to the results of Dupret et Zhu and Dupret et al in which the Phyllolepida are a different actinolepid branch than the one leading to the Phlyctaenioidei. This different result must be linked to the lack of taxa of the relevant subgroups (Actinolepidoidei, including Wuttagoonaspididae and Antarctaspididae; Phlyctaenii). Node11 to Gavinaspis (Parietals (preorbitals of placoderms) do not surround pineal foramen or prominence; CI = (Posterior process of the paranuchal plate behind the nuchal plate (dorsal face) absent; CI = 0.500) 10

11 It is also noteworthy that the centronuchal plate of Gavinaspis (and thus of the Phyllolepida) has been considered as a single central plate, rather than a nuchal plate or a fusion of central and nuchal plates (details are given in Dupret et Zhu, ). Node 11 to Cowralepis 20 1 (Extent of dermatocranial cover incomplete (scale-free and elsewhere; CI = 0.200) Node 10 to node 12 (Phlyctaenioidei; present in 50% of the trees) (dermal neck-joint of ginglymoid type (phlyctaenioid type) with dermal shoulder girdle plate with articular condyle or fossa;ci = 0.800) (Middle and posterior pit-lines on postparietal posteriorly situated; CI = 0.250) The uncertainty of this clade is to be referred to the resolution of Dicksonosteus. The latter is the only representative of the Phlyctaenii group (because best known owing to Goujet ); but it should be kept in mind that some other Phlyctaenii are more basal (Phlyctaeniidae, Groenlandaspididae; works in progress). Node 12 to Dicksonosteus 34 0 (Shape of opercular (submarginal) ossification is a broad plate that tapers towards its proximal end; CI = 0.333) 47 0 (Insertion area for jaw adductor muscles on palatoquadrate ventral; CI = 0.500) Node 12 to node 13 (Brachythoraci) 76 0 (Jugular vein invested in otic capsule wall posterior to the postorbital process; CI = 0.500) (Pronounced internal crista (keel) on median dorsal surface of shoulder girdle present; CI = 1) (Posterior end of supraorbital canal in postparietal; CI = 0.444) Node 13 to Coccosteus 17 0 (Jugal portion of infraorbital canal joins supramaxillary canal present; CI = 0.333) Node 8 to node 14 (unnamed clade) 5 2 (Dentine kind: orthodentine; CI = 0.400) 12 0 (Body scales with bulging base absent; CI = 0.167) 13 1 (Body scales with flattened base absent; CI = 0.167) 21 1 (Endolymphatic ducts open in dermal skull roof absent; CI = 0.500) 55 0 (Median dermal bone of palate (parasphenoid) absent; CI = 0.333) 60 1 (Position of myodome for superior oblique eye muscles anterior and dorsal to 11

12 foramen; CI = 1) 74 1 (Basicranial morphology tropibasic; CI = 0.250) 82 1 (Labyrinth cavity with skeletal capsular wall absent; CI = 1) 88 1 (Trigemino-facial recess present; CI = 1) 89 1 (Posterior dorsal fontanelle present; CI = 0.500) (Dorsal fin spines present; CI = 0.200) (two dorsal fins present; CI = ) (Caudal radials restricted to axial lobe; CI = 0.500) (Middle and posterior pit-lines on postparietal posteriorly situated; CI = 0.250) (Position of middle and posterior pit lines close to midline; CI = 0.200) (Metapterygoid with developed medial ventral protrusion present; CI = 0.500) (Shape of parasphenoid denticulated field is broad, splint-shaped slender, splintshaped; CI = 0.500) (optic fissure absent; CI = 0.500) Node 14 to Entelognathus 27 1 (Bony hyoidean gill-cover series (branchiostegals present; CI = 0.167) 35 1 (Gular plates present; CI = 0.500) 45 1 (Maxillary and dentary tooth-bearing bones present; CI = 0.500) (Ventral margin of maxilla curved; CI = 0.50) (Posterior end of in nuchal plate; ancestral state was "1", in postparietal; CI ) (infradentary present; CI = 0.500) Contrary to Zhu et al. s (2013) analysis 12, Entelognathus appears in a resolved position, at the top of the placoderm grade, but still more basal than Ptyctodonts (most likely because of its basal arthrodire and actinolepidoideid traits). This is most likely due to the change in the coding of some characters (better illustrating the diversity of placoderms like the type of dermal craniothoracic neck joint). Node 14 to node 15 (unnamed clade) 24 0 (Consolidated cheek plate absent; CI = 0.250) 72 1 (Short otico-occipital re region of braincase present; CI = 0.250) 75 1 (Ascending basisphenoid pillar pierced by common internal carotid present; CI = 0.500) 94 0 (Process forming part or complete wall of jugular groove or canal projecting from otic capsule wall absent; CI = 0.500) 96 1 (Ventral cranial fissure present; CI = 0.333) 97 1 (Metotic (otic-occipital) fissure present; CI = 0.500) (Pectoral fenestra completely encircled by dermal shoulder armour; CI = 0.333) (Junction of posterior pitline and main lateral line close to posterior margin of 12

13 skull roof; CI = 0.500) (Type of dermal neck- joint is spoon-like(ptyctodont and petalichthyid like); CI = 0.800) (More than four sclerotic plates; CI = 0.500) (Median gular present; CI = 0.333) (Contact between palatoquadrate and dermal cheek bones metapterygoid and autopalatine contacts separated by gap between commissural lamina of palatoquadrate and cheek bones; CI = (Developed postorbital cavity absent; CI = 0.500) (Posterior postorbital process absent; CI = 1) Only two characters (215 and 230) show a consistency index of 1, but none of them consists in a non ambiguous transformation. Node 15 to node 16 (Ptyctodontida) 16 1 (Sensory line network passes through canals enclosed within dermal bones; CI = 0.250) 23 0 (Series of paired median skull roofing bones that meet at the dorsal midline of the skull (rectilinear skull roof pattern absent; CI = 0.333) 68 1 (Narrow interorbital septum present; CI = 0.200) 95 0 (Position of hyomandibula articulation on neurocranium below or anterior to orbit, on ventrolateral angle of braincase; CI = 0.500) (Pelvic claspers present; CI = 0.333) (Dermal pelvic clasper ossifications present; CI = 1) (One pair of paranuchal plates; CI = 0.333) (Large unpaired median bone contributing to posterior margin of skull roof absent; CI = 0.500) (Posterior end of supraorbital canal in nuchal plate; CI = 0.444) (Metapterygoid with developed medial ventral protrusion absent; CI = 0.500) (Pelvic fin spine absent; CI = 0.250) Node 16 to node (Shape of opercular (submarginal) ossification narrow, rod-shaped; CI = 0.333) Node 17 to Austroptyctodus (Endoskeletal postbranchial lamina on scapular process present; CI = 0.250) 13

14 Node 17 to Rhamphodopsis (Fin spines with ridges present; CI = 0.333) Node 15 to node 18 (Eugnathostomata) 7 0 (Lepidotrichia or lepidotrichia-like scale alignment present; CI = 0.200) 14 1 (Flank scales alignment in oblique rows or hexagonal/rhombic packing; CI = 0.500) 37 1 (Basihyal absent, hyoid arch articulates directly with basibranchial; CI = 0.500) 46 1 (Large otic process of the palatoquadrate present; CI = 0.250) 67 1 (Extended prehypophysial portion of sphenoid present; CI = 0.250) (Dermal shoulder girdle forming a complete ring around the trunk present; CI = 1) (Median dorsal plate absent; CI = 0.500) (Fin spines with ridges present; CI = 0.333) (Synarcual absent; CI = 0.500) (Anal fin present; CI = 0.333) (One single marginal bone alongside paired median skull roofing bones over the otico-occipital division of braincase; CI = 0.333) (Dermal neck-joint between paired main-lateral-line-bearing bones of skull and shoulder girdle absent; CI = 1) (Opercular suspension on braincase present; CI = 1) (Presupracleithrum present; CI = 0.500) (Pelvic girdle without substantial dermal component; CI = 0.333) Node 18 to node 19 (unnamed clade Chondrichthyes + Acanthodii) 2 1 (Perichondral bone absent; CI = 0.167) 11 0 (Body scale profile distinct crown and base demarcated by a constriction (neck); CI = 0.200) 18 1 (Dermal skull roof consists of undifferentiated plates or tesserae; CI = 0.500) 21 0 (Endolymphatic ducts open in dermal skull roof present; CI = 0.50) 28 0 (Branchiostegal plate series along ventral margin of lower jaw absent; CI = 0.333) 33 0 (Opercular (submarginal) ossification absent; CI = 0.500) 40 1 (Tooth whorls present; CI = 0.125) 44 0 (Dermal jaw plates on biting surface of jaw cartilages absent; CI = 0.333) 52 1 (Preglenoid process present; CI = 0.500) 54 1 (Precerebral fontanelle present; CI = 0.500) 71 1 (Glossopharyngeal nerve (N. IX) exits through metotic fissure; CI = 1) 81 1 (Postorbital process articulates with palatoquadrate present; CI = 0.500) 91 1 (Dorsal ridge present; CI = 0.500) (Macromeric dermal shoulder girdle absent; CI = 0.250) (Dermal shoulder girdle girdle with ventral components only; CI = 0.500) (Median fin spine insertion deep; CI = 0.125) 14

15 138 0 (Caudal radials extend beyond level of body wall and deep into hypochordal lobe; CI = 0.500) (Hyoid arch articulation on otic capsule wall; CI = 1) Node 19 to node 20 (Acanthodii) 5 0 (Dentine kind mesodentine; ancestral state was orthodentine; CI = 0.400) 68 1 (Narrow interorbital septum present; CI = 0.200) 94 1 (Process forming part or complete wall of jugular groove or canal projecting from otic capsule wall present; CI = 0.500) (Anal fin spine present; CI = 1) (Paired fin spines present; CI = 0.500) (Intermediate fin spines present; CI = 0.167) (Prepectoral fin spines present; CI = 0.250) (Pelvic fin spine present; CI = 0.250) Node 19 to node 39 (Chondrichthyes) 1 1 (Tessellate prismatic calcified cartilage present; CI = 1) 7 1 (Lepidotrichia or lepidotrichia-like scale alignment absent; CI = 0.200) 14 2 (Flank scales alignment disorganised; CI = 0.500) 20 1 (Extent of dermatocranial cover incomplete (scale-free and elsewhere); CI = 0.200) 37 0 (Basihyal present; CI = 0.500) 41 1 (Bases of tooth whorls with some or all whorls consist of separate tooth units; CI = 0.500) 74 0 (Basicranial morphology platybasic; CI = 0.250) 75 0 (Ascending basisphenoid pillar pierced by common internal carotid absent; CI = 0.500) 77 1 (Canal for lateral dorsal aorta within basicranial cartilage present; CI = 0.250) 79 1 (Canal for efferent pseudobranchial artery within basicranial cartilage present; CI = 0.500) 93 1 (Lateral otic process present; CI = 0.333) (Scapular process with posterodorsal angle present (CI = 0.500) (Pelvic claspers present; CI = 0.333) Node 18 to node 50 (Osteichthyes) 3 1 (Extensive endochondral ossification present; CI = 0.500) 10 1 (Body scales with peg- and-socket articulation present; CI = 0.333) 15 2 (Sensory line canal perforates and passes through scales; CI = 0.400) 27 1 (Bony hyoidean gill-cover series (branchiostegals) present; CI = 0.167) 35 1 (Gular plates present; CI = 0.500) 38 1 (Interhyal present; CI = 1) 15

16 43 0 (Teeth ankylosed to dermal bones absent; CI = 1) 45 1 (Maxillary and dentary tooth-bearing bones present; CI = 0.500) 48 1 (Oblique ridge or groove along medial face of palatoquadrate present; CI = 0.250) 55 1 (Median dermal bone of palate (parasphenoid) present; CI = 0.333) 62 1 (Spiracular groove on basicranial surface present; CI = 0.333) 63 1 (Spiracular groove on lateral commissure present; CI = 0500) 87 1 (External (horizontal) semicircular canal joins level with posterior ampulla; CI = 0.500) 98 1 (Vestibular fontanelle present; CI = 0.500) (One or two dorsoventrally offset spino-occipital nerve foramina; CI = 0.500) (Parachordal shape keeled with sloping lateral margins; CI = 0.500) (Scapular process of shoulder endoskeleton absent; CI = 0.500) (Coracoid process absent; CI = 0.333) (Fin base articulation on scapulocoracoid eurybasal; CI = 0.250) (Dorsal fin spines absent; CI = 0.200) (infradentary present; CI = 0.500) (Dermal plates on mesial (lingual) surfaces of Meckels cartilage and palatoquadrate present; CI = 1) (Biconcave glenoid on lower jaw present; CI = 1) (Fenestra ventrolateralis absent; ancestral state was "2" common ventral fenestra for anterior and posterior nostrils; CI = 0.500) (Postorbital pila ascending from basipterygoid process to postorbital process present; CI = 0.500) (Lateral cranial canal present; CI = 0.500) 16

17 Data matrix (modified from Zhu et al, ) Osteostraci ?010?0? ? ? ? ??00-00-?00?-000-?? ??--? ? ?? ? ? ? ? ?0??0-??---?0??? ???0--01 Galeaspida ??00101?0?1000? ? ?0-? ? ??01-00-?00? ?--? ?--??--?----0??? ? ? ? ?0???-??---?0??? Parayunnanolepis 0??1?0?1?0?10? ????????0??????????00????????0?????????????????????????????????????000010?????????0? ?00---?00?????----?1???0?0? ?-1???-??? ?-???????0----? ??---????--?0???????????????????????0?0????00010 Bothriolepis ?--?--? ???---?10000??00?0?000???????0?????00?????????????????????0???????? ????? ??0??----?1???0? ?- 12?0-??? ?-????3??0----? ?---0?00--?0???????????????????????0?0????0--10 Pterichthyodes 0?0??0-1?0??? ??10-0? ??0--???0??????????000???????0????????????????????????????0????????0?0010????? ?00? ?????----?1???0? ?- 12??-??? ?-????3??0----? ??---????--?????????????????????????0?0???????10 Brindabellaspis ?100010????0-?000?1-?????010???????????????????0? * ?00?? ?0000???? ??????0??0??0---???????0--??01???0? ??-???????????-??? ?????????????????-????--02?100????11??00?????????????0?????

18 Macropetalichthys 00?110?????????1?0-?00001-???????0???????????????????0011?0?000000? ?1-1?0???10?00-? ?00?00??????????????????????????0--???????0--??00????? ?3??-????------?-???030?0---?? ??---????--???100????110000?????????0?0???????1? Romundina 00010/10????????? ?????010?????0--??000?????? ? ??101?0?????00?0-?00--?????000???? ? ??3? ?040?0--1??- 0???????????????00?? ?0???? ??10 Sigaspis 0????0??? ? ?-?????010??????????????????????????????01?????????????????????????????????????000010?????????1?1?0000?0?000--??0??????????0?????0?00? ?10?0-??? ?-11??10?0--??? ??---????--?????????????????????????0?0?????001? Gavinaspis?00??0?????????0?0-0010??-?????????????????????????????1????????01???????0?????????????????00??????0?????????????????????????????????????????????????????????00?10110??10??????????????-10??10?0--???????????????????????????????????????????????????????????10 Cowralepis 000??0?????????0?0-1?10-1-0????0110-0? ?????00??1????????????????????????????????????0??????????? ????????01?1? ????----?0?????0-00? ??-????------?-???010?0----? ?---00??--?????????????????????????0?0?????101? Dicksonosteus 00?110?1?0????? ??1?? ???? ?00?? ? ? ????000?0-?00--1??????0--??00????? ??? ?-0/11? ? ?00000???10 Coccosteus 000??01???????? ?? ??????00?11?1???????1?????00??????????????????0??????????? ?0?--1?????1? ?????----?0?????

19 11?0-??? ?-01?000?0----? ?---000?--?????????????????????????0?0?????1010 Buchanosteus 00?1?0?100010?? ??1-- 0?10010??0? ?00??1000??? ?0?--1?1???????00?0??00--???????0--??00????? ??-????------?-????0??0----? ? ?? ?00??????10 Austroptyctodus 0001?0?1?01010?1?0-? ????000?01010?0????101????????0?????????????????0???0?? ? ?00--11??????--??00?????0-00? ? ?-???03??0----? ??---0??0--??????????1??????????????0?0???????1? Campbellodus 000??011?01010?1?0-? ?0??000?0??????????1????????????????????????????0???????? ???--10?0?? ?00--1?01????----?0?????0-00? ? ?-???03??0----? ?? ??????????1??????????????0?0?????0?1? Rhamphodopsis 000??0???????0?1?0-? ?? ????????0??????????1????????????????????????????0????????0 0011??0?--10???? ?01--?10?????--???0?????0-00?1010??01-13? ?-???03??0----? ??---0???--??????????1??????????????0?0??????11? Entelognathus 000??0???0101??0? ???1?????101?????00?1010? ????00?01????0???????????011000??10?000010??????????????00??0????--????????--??00????00-00?? ?10000???0000?1???????2??00??0?0?????110?0??????01??0???1??0? 01100?00????????0??0??????1? Acanthodes ?-0? ? ?01?1??010?0111? ? ??? ?-----? ? ???? ?-0? ?????0?001?00?? ?

20 Achoania ?????????1?0-????10??????????????1?1?11??????0?011010?1??00100???1?????0?????????????????1???????00?1???????????1????????0???????????0?0????1???110???01?????????????0????100???????1010???1???1110?? ???? ?????????010?0010??????01??11 Akmonistion ?0-??21-?1? ? ?000? ? ?111?11?001?? ? ?0?? ?-----? ?---? ?????---? ?-??---0? ?0????1?00? ??01?011 Brachyacanthus 0?0??00??????10??100??-0? ????--- 00??????????1?1????????????????????????????????????????????? ?0??1? ?????--??-----?-----? ?---?--?--????----?--?????---?-????---0-?-???--????--????????????????---?-??------?????11? Brochoadmones ? ??-0? ?? ??????????1?1????????????????????????????????????????????? ??????-?-?? ?????--??-----?-----? ?---?--?--????----?--?????---?-????---0-?-???--????--????????????????---?-??------??????1? Cassidiceps 000??00??00101?0?10???-???111?100-0-?? ???????????1?1????????????????????????????????????????????? ?????????1? ?????--??-----?-----? ?---?--?--????----?--?????---?-????---0-?-???--????--????????????????---?-??------??????1? Cheiracanthus ?111??--? ?? ?01?1?????????????????????01?????????????????????? ?1?? ?????--??-----?-----? ? ?--????----?--?????---?-????---0-?-???--???1--????????????????---?-??------?????11? Cheirolepis 20

21 ?? ?10000?11?10??????11????10????0?0??00???????????1??????? ?????1??? ?000?--11? ? /10000??0? ?00?111100?0??????0??????10?000101?0/10101? Chondrenchelys ?00??-1-?1? ?? ? ?0?000110??-111??0?????????????0?00110?0? ?????--??-----?-----? ?---?--?--????----?--?????---?-????---0-?-???--???1--????????????????---?-??------??????11 Cladodoides 110????????????-????---?-?????????????111?- 0011?00???1010??? ? ???????????????????????????????????????--??-----?-----? ?---?--?--???? ?????---??????---0-?-???--0? ?100???1?0??00---?-??--?---??????11 Cladoselache ??2?-?1? ? ?0?01?10?0??0111????10???11?11???????1?1?101010??10? ?000000????--??-----?-----? ?---?--?--????----?--?????---?-0???---0-?-???--???1--?????????1??????---?-??------????1011 Climatius ?? ? ? ?? ?00010?0??1????????????????????????????????????????????? ?0??1? ????? ?-----? ? ?--? ?????---?-?? ?-???--00?1--???????????????? ?---?????11? Cobelodus 1101?01??0-????-?1? ? ?1?? ????--??-----?-----? ?---?--?--????----?--?????---?-????---0-?-???--???1-- 02?1?0?0????0?00---?-??------??????1? Culmacanthus?00??00?? ?100??-1? ????---?0????????????1?????????????????????????????????????????????

22 10-10?0010?0??10? ????? ? ?---?--?--????----?--?????---?-????---0-?-???--????--????????????????---?-??------?????11? Debeerius ?0???-1-?1? ?- 0001?000? ?0?0?0111??-111??0?110?????? ?0?00? ? ????--??-----?-----? ?---?--?--????----?--?????---?-????---0-?-???--???1--????????????????---?-??------???01?1? Diabolepis ?????????1?0-?1-1??????????????? ?????000011?10?0?100110???101?1?0?0?1??11??????00111??????????????????????????????????????????001????0? ?00101?????? 0-??1????000??? ?11??-? ?10????? ??0?0?0?0?? ??????????11 Dialipina 0????00?011011?0? ??????0??10??1?- 00?1??????????1?1??????????????0?????????????????????????????? ??????????? ??11?0??????1??10??00/1?00?110???-??0-?0????????10????01010?0???1?10????11??0????????1????????????????????????????00?????01? Diplacanthus 0101? ?0100??-0? ??0-?? ?????1??????1????????????????????????????????????????????? ?? ????? ?-----? ? ?--? ?????---?-?? ?-???--0???--????????????????---?-??--?---?????11? Doliodus 1?0??0?100?????-?1??---0-?0----?0-0-?? ??????101010?00?00010???00??111?11?00????11?? ?0???? -????????????????00??1???11????????? ?-----? ?---?--?--? ?????---?-?? ?-???--0? ??0????1?0?? ?---????1?11 Eusthenopteron ? ? ? ?