The oldest three-dimensionally preserved vertebrate neurocranium

The neurocranium is an integral part of the vertebrate head, itself a major evolutionary innovation1,2. However, its early history remains poorly understood, with great dissimilarity in form between the two living vertebrate groups: gnathostomes (jawed vertebrates) and cyclostomes (hagfishes and lampreys)2,3. The 100 Myr gap separating the Cambrian appearance of vertebrates4–6 from the earliest three-dimensionally preserved vertebrate neurocrania7 further obscures the origins of modern states. Here we use computed tomography to describe the cranial anatomy of an Ordovician stem-group gnathostome: Eriptychius americanus from the Harding Sandstone of Colorado, USA8. A fossilized head of Eriptychius preserves a symmetrical set of cartilages that we interpret as the preorbital neurocranium, enclosing the fronts of laterally placed orbits, terminally located mouth, olfactory bulbs and pineal organ. This suggests that, in the earliest gnathostomes, the neurocranium filled out the space between the dermal skeleton and brain, like in galeaspids, osteostracans and placoderms and unlike in cyclostomes2. However, these cartilages are not fused into a single neurocranial unit, suggesting that this is a derived gnathostome trait. Eriptychius fills a major temporal and phylogenetic gap in our understanding of the evolution of the gnathostome head, revealing a neurocranium with an anatomy unlike that of any previously described vertebrate.

Colorado, form the syntype USNM V 2350 in the collections of the Smithsonian National Museum of Natural History in Washington, DC, USA.Emended diagnosis.Agnathan with mesomeric dermal tesserae and scales formed from acellular bone overlain by ornament formed from coarse wide-calibre tubular dentine.Body scales covered in multiple elongate ridges.Antorbital neurocranium comprising symmetrical set of elements containing numerous large canals internally, endoskeleton closely associated with but not fused to the surrounding dermal skeleton.Shared with Astraspis, arandaspids, other 'ostracoderms' excluding heterostracans: multiple branchial openings.Shared with Astraspis and tessellate heterostracans: dermal head skeleton formed from dorsal and ventral 'headshields' of ornamented tesserae.Differs from Astraspis in that ornament of dermal skeleton comprises ridges and presence of coarse tubular dentine.Description.Computed tomography scanning of the part and counterpart of PF 1795 (here termed PF 1795a and b, respectively; Methods) confirms the identity of this material as a partially articulated Erip tychius head 8 , including components of both the dermal skeleton and endoskeleton (Fig. 1, Extended Data Figs.1-3 and Supplementary Video; see Supplementary Information for comments on histology).The articulated individual is confined to the near surface of the matrix; below it is a mash of isolated elements typical of the Harding Sandstone bone beds including additional tesserae referable to Eriptychius that do not seem to be associated with the articulated specimen (Extended Data Fig. 1).Denison 8 described several large elements of globular calcified cartilage as part of the internal skeleton of Eriptychius and we have been able to distinguish ten separate cartilages in total comprising the endoskeletal cranium (Figs. 1 and 2 and Extended Data Figs. 2 and 4).
Six cartilages were identified on the split surface of PF 1795a by ref. 8 (Extended Data Fig. 2) and the remaining four are buried within the matrix of PF 1795b.The cartilages are closely wrapped by articulated squamation anteriorly, dorsally and ventrally and to one side (Extended Data Figs. 3 and 4); however, there is a clear separation between dermal and endoskeletal elements, unlike in galeaspids and osteostracans 36 .This squamation comprises a range of dermal element types including the scale types identified by ref. 8 and scales with an anteroposteriorly oriented ornament from farther back on the head.It also includes small, curved orbital plates (Fig. 1c,d and Extended Data Fig. 3c-e) and several plates similar in morphology (Fig. 1c,d and Extended Data Fig. 3f-i

Article
Denison identified two cartilages as the orbital cartilages on the basis of concave posterior faces 8 .Our scan data confirm that these represent cross-sections through large fossae that we interpret as forming the anterior wall of the orbits.These fossae are flanked laterally by an antorbital process (Fig. 2a,b and Extended Data Fig. 5) that suggests a dorsolateral orientation of the orbit.A smaller ventral fossa on each orbital cartilage below the orbit may have provided a location for muscle attachment (Fig. 2).One orbital cartilage is posteriorly displaced, along with elements of anterior squamation including a rostral scale (Fig. 1 and Extended Data Fig. 3) and has rotated by about 180° in the sagittal axis; when rotated back into position it aligns with the orbital plates (Fig. 1).The anteriormost branchial plate lies posterolateral to the other orbital cartilage (Fig. 1 and Extended Data Fig. 3), suggesting that the relative positions of the orbits, otic region and branchial arches were similar to those inferred in heterostracans 2,19 , although this is impossible to judge exactly because of the collapse of the dermal skeleton.The absence of anything assignable to the branchial skeleton suggests that the branchial arches were not incorporated into a single mineralization with the neurocranium, a major difference with respect to galeaspids and osteostracans 7,15 .
The remaining cartilages lie between the two orbital cartilages, although they have slumped slightly and been pulled posteriorly on one side with the displaced orbital cartilage (Figs. 1 and 2a-c and Extended Data Fig. 4).There are three paired cartilages arranged symmetrically across the midline-two dorsal (termed mediolateral A, B) and one ventral (termed mediolateral C)-along with one unpaired midline cartilage dorsally and one ventrally.Of the two unpaired cartilages, the smaller dorsal element is kite-shaped and has both the dorsal and ventral surfaces punctured by large medial foramina that we consider likely to be the pineal opening (Fig. 2a,e).These dorsal and ventral foramina do not exactly line up anteroposteriorly but appear to communicate with a common large space inside the cartilage (Fig. 2g and Extended Data Fig. 6).This kite-shaped cartilage is preserved overlying the mediolateral cartilages A and its ventral surface bears a median ridge with a shallow depression on either side (Fig. 2a,e).Together with shallow depressions on the dorsal surface of mediolateral cartilages A, these depressions frame paired fossae which we interpret as having accommodated part of the forebrain, possibly the olfactory bulbs (Fig. 2d-f), an interpretation consistent with their position relative to the orbits and putative pineal opening.Thus, we infer this to be the dorsal side of the animal and term this the median dorsal cartilage.The larger median ventral cartilage underlies the mediolateral cartilages.
All cartilages are pervasively penetrated by canals (Fig. 2g and Extended Data Fig. 6).In the larger cartilages, that is, the orbital cartilages and the mediolateral cartilages A, this tends to follow the pattern of a larger trunk entering the cartilage from the posterior side before splitting into smaller branches that open to the surface.The pattern is not exactly bilaterally symmetrical in the paired and unpaired elements but does follow a similar organization with the trunk canal entering at equivalent points.These canals could plausibly have carried sensory rami onto the surface of the head; for example, the superficial   ophthalmic nerve in the case of the canals in the orbital cartilages.However, the canal openings are not restricted to surfaces where the cartilages contact the dermal skeleton.On the basis of this and the lack of sensory canal openings in the head tesserae, they may have carried vasculature instead or as well.In living chondrichthyans, canals carry vasculature and transport prechondrocytes into the cartilage from the perichondral surface.This could implicate the canals in Eriptychius in both cartilage maintenance and interstitial cartilage growth, although in modern chondrichthyans the width of these canals are an order of magnitude smaller than in Eriptychius 38,39 .Although canals have not been explicitly reported in other early vertebrate cartilages, sections through galeaspid cartilage suggest that this tissue has a degree of vascularity 36 .
In concert with the displacement of the cartilages, the dermal squamation has undergone postmortem collapse.We interpret the articulated area of squamation as having covered the right side of the head, around the region of the right orbit and the right side of the mouth, as well as the areas dorsal and ventral to the mouth (Fig. 1 and Extended Data Fig. 3).On the basis of this interpretation, the cartilages would have comprised the preorbital region of the head, surrounding the mouth (Fig. 2c,h,i).The articulated dorsal and ventral patches of squamation indicate that the mouth opened between the cartilages, bordered by the 'rostral' plates 8 and was oriented supraterminally (Fig. 2c,h,i), unlike most jawless stem-gnathostomes (notable exceptions being Doryaspis 40 and Drepanaspis 41 ).The fact that the cartilages are separate may mean that some movement of the oral endoskeleton was possible, although this would have presumably been limited by their close relationship with the squamation.

The evolution of vertebrate crania
The cranial cartilages of Eriptychius have no obvious homologue in the head of any known extant or extinct vertebrate (Fig. 2 and Extended Data Fig. 4).The most obvious comparison is with the numerous paired and midline cartilages that comprise the complex feeding apparatus of cyclostomes 42,43 and possibly anaspid-like early vertebrates 17 .The preorbital region of Eriptychius is ostensibly similar in that it comprises numerous cartilages that bordered the mouth.Unlike cyclostomes, however, in which the entire brain is held in an open cartilaginous framework 10 , in Eriptychius the forebrain at least, as well as the orbits, were bounded by these cartilages.The only large cartilage that is not associated with the forebrain or orbit, the median ventral cartilage, is closely associated with the dermal skeleton, suggesting that mobility would have been restricted.Although it is impossible to rule out movement of the smaller mediolateral cartilages B and C, a cyclostome-like mobile feeding apparatus seems unlikely.
The preserved cranial cartilages of Eriptychius instead seem to have formed a static structure, which, in this sense, is more comparable to neurocrania in known jawless stem-group gnathostomes.In these taxa, osteostracans and galeaspids, endoskeletal tissues fill the connective tissue space between the dermal skeleton and the brain, where they surround the mouth and pharynx and buttress the head 1,7,14 .The cartilages in Eriptychius are similar in that they fill out the head and closely support the dermal skeleton.Unlike osteostracans and galeaspids, however, the neurocranial cartilages are separate from one another and there is no evidence for any mineralization from the level of the orbits posteriorly or for the fusion of the splanchnocranium into a unit with the neurocranium.This could be taphonomic but parts of the dermal skeleton posterior to the orbits remain articulated in PF 1795.Instead, Eriptychius may have resembled heterostracans, in which the otic region of the brain and the pharynx are closely associated with the dermal skull roof 19,44 , suggesting that they were primarily supported by the dermal skeleton.The patterning of developmental cartilages in extant gnathostomes might suggest that the mineralization of this anterior region in Eriptychius was limited to the prechordal region of

Article
the neurocranium (the trabeculae cranii), which originate from neural crest 3,12 .However, it has been demonstrated that in lampreys the parachordals extend forward to form most of this region 3,12 , another possible origin for the cartilages in Eriptychius.
Eriptychius fills an important gap, both temporal and phylogenetic, in our understanding of the evolution of the vertebrate head.Our inclusion of Eriptychius in a recent phylogenetic matrix for early vertebrates 45,46 recovers it within or in a polytomy with the vertebrate crown group and as a stem-group gnathostome in the Adams consensus of the parsimony analysis (Fig. 3 and Extended Data Figs.7-9) which is consistent with previous phylogenetic analysis 35 .This phylogenetic placement would extend the condition in which endoskeletal tissues fill space in the neurocranium into the earliest, Ordovician, stem-group gnathostomes and show that this is not limited to taxa with an osteostracan/galeaspid morphology of a ventrally positioned mouth and dorsally located orbits.However, the fact that in Eriptychius the cartilages are not fused into a single unit around the brain suggests that these early gnathostome neurocrania calcified in several parts.An enclosing neurocranium, which in galeaspids and osteostracans is fused with the splanchnocranium, is a trait which unites galeaspids, osteostracans and mandibulate gnathostomes to the exclusion of Eriptychius and cyclostomes.Although it is possible to identify broad similarities, the substantial difference between the neurocranial anatomy of the Ordovician Eriptychius to either cyclostomes or gnathostomes helps to explain why it has been so difficult to draw direct anatomical comparison between the skulls of the two living vertebrate groups 2 .
Eriptychius provides the earliest direct evidence for a prechordal endocranium in a vertebrate.This was also likely to have been the case in the contemporary Astraspis and the later heterostracans based on the lateral positions of the orbits 33,47,48 (Fig. 3d,e).This contrasts with Sacabambaspis, in which the orbits are placed at the extreme anterior margin of the headshield 30 (Fig. 3c) comparable to those of putative stem-group vertebrates 4 such as Haikouichthys, Metaspriggina and conodonts [4][5][6]49 (Fig. 3a) as well as in the naked anaspids Jamoytius and Euphanerops 50,51 . Inthe past, this unusual anatomy has usually been dismissed as a specialization on the basis of interpretations of Saca bambaspis in a heterostracan light (for example, refs.27,31 and supplementary appendix (p35) of ref. 45).The discovery of this preorbital neurocranium in Eriptychius and the movement of early vertebrate taxa around the vertebrate crown node in recent phylogenies 45,46 should prompt reconsideration of whether differences in orbital placement in Ordovician vertebrates instead reflect a more fundamental anatomical difference in cranial organization.

Online content
Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41586-023-06538-y.Specimen E. americanus PF 1795 is held at the Field Museum, Chicago, USA.It comprises two pieces, part and counterpart, one of which had the split face set in epoxy and was manually prepared out of the matrix from the other side 8 .Denison 8 figured only the part in epoxy (before and after preparation) and referred to both part and counterpart as PF 1795: here, we term the part in epoxy PF 1795a and the unprepared counterpart PF 1795b for the sake of clarity (Fig. 1).

Geological setting
The Harding Sandstone (Sandbian, Upper Ordovician) is a thin unit of interbedded mudstones and sandstones that extends around the Cañon City embayment in the frontal range of the Rockies.The sequence was first studied in detail by ref. 54 focusing on the presence of Ordovician vertebrates with periodic attention paid subsequently.Sedimentologically, the sequence records shallow marine deposition within a microtidal lagoonal setting with localized estuarine input 55 .The extensive bone beds that occur through the Harding represent winnowed shoreface deposition with specimen PF 1795 described here thought to come from an equivalent or potentially the same horizon to the articulated specimen of Astraspis desiderata that represents a shoreface strandline 33 .

Scanning
Computed tomography was carried out at the University of Chicago on a Phoenix v|tome|x with a dual 180 tube.PF 1795a was scanned at a voltage of 100 kV and current of 370 μA with a 0.1 mm Cu filter, for 1,800 projections, achieving a voxel size of 34.17 μm.PF 1795b in the matrix was scanned at a voltage of 110 kV and current of 300 μA with no filter, for 2,000 projections, achieving a voxel size of 44.8940 μm.Both datasets were segmented in Mimics v.25 (materialize) to create three-dimensional meshes using manual segmentation with some interpolatory functions ('3D interpolate' and 'Multiple Slice Edit').These were exported in the PLY format and then visualized in Blender (blender.org)v.3.3.0.An additional, higher resolution scan of PF 1795a was carried out at the University of Bristol in an effort to better visualize the endoskeletal tissue; this was carried out at a voltage of 120 kV and current of 119 μA with no filter obtaining 3,141 projections with a voxel size of 14.72 μm.

Phylogenetic analysis
The phylogenetic analysis was conducted on the basis of the matrix of ref. 45, with minor modifications focussed on Ordovician vertebrates.Eriptychius was added, astraspids recoded as Astraspis and arandaspids as Sacabambaspis.We also revised the codings of heterostracan taxa, revising Athenaegis and splitting Heterostraci into Anglaspis and Erri vaspis.Changes are listed and justified in the Supplementary Information.This resulted in a matrix including 54 taxa and 167 morphological characters, which we analysed using parsimony and Bayesian analyses.
Parsimony analysis was carried out in TNT v.1.5using a parsimony ratchet and TBR branch swapping with 10,000 replicates, holding 100 trees from each iteration, with the constraint (Hemichordata (Cephalochordata (Tunicata + all other taxa))) and Hemichordata set as the outgroup.This resulted in 1,951 equally parsimonious trees with a length of 351.Bayesian analysis was carried out in MrBayes v.3.2.7, a flat (uniform) prior was used with an Mkv model and a gamma-distributed rate parameter.Hemichordata was set as the outgroup and total-group vertebrates were constrained to be monophyletic.We carried out the search for 10,000,000 generations, sampling a tree every 1,000 generations and calculated a majority rule consensus tree with a relative burn-in of 25%.

Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.

n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one-or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section.

A description of all covariates tested
A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g.means) or other basic estimates (e.g.regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g.confidence intervals) For null hypothesis testing, the test statistic (e.g.F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted Give P values as exact values whenever suitable.
For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes Estimates of effect sizes (e.g.Cohen's d, Pearson's r), indicating how they were calculated Our web collection on statistics for biologists contains articles on many of the points above.

Software and code
Policy information about availability of computer code Data collection not applicable

Data analysis
For analysis of the image data we used the paid software Mimics v.25 (Materialise) to make 3D models and the free software Blender (blender.org)to acquire images.For the phylogenetic analysis we used the software tnt and MrBayes.
For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and reviewers.We strongly encourage code deposition in a community repository (e.g.GitHub).See the Nature Portfolio guidelines for submitting code & software for further information.

Specimen provenance
The specimen is held in the Field Museum, Chicago, USA and was taken out on loan for the duration of the study.

Specimen deposition
The specimen continues to be held at the Field Museum

Dating methods
No new dates are provided Tick this box to confirm that the raw and calibrated dates are available in the paper or in Supplementary Information.

Ethics oversight
None required, material already held in collections Note that full information on the approval of the study protocol must also be provided in the manuscript.

Fig. 1 |
Fig. 1 | Overview of Eriptychius americanus PF 1795.a,b, Photographs of part PF 1795a, which had the split face set in epoxy and was manually prepared (a), and its counterpart PF 1795b, which remains in rock matrix (b).Both are shown in an anatomically ventral view.c,d, Digital model of computed tomographic data of the combined part and counterpart with most of dermal skeleton rendered transparent: anatomical ventral view (corresponding to the visible area of the part in epoxy) (c) and anatomical dorsal view (buried in matrix in the counterpart) (d).Colour scheme for renders: blue-greys, cranial cartilages (matching the detailed scheme in Fig.2); transparencies, the dermal skeleton; orange, branchial plates; red, orbital plates.Anterior to top in a-d.ant.tess., anterior tesserae; artic.vent.tess., articulated ventral tesserae; branch.plate, branchial plate; cran.cart., cranial cartilages; disp., displaced; frag., fragment; L., left; orb.cart., orbital cartilage; orb.plates, orbital plates; R., right; vasc., vasculature; ?, probable.Scale bar applies to all panels.

Fig. 2 |
Fig. 2 | The neurocranial cartilages of E. americanus PF 1795 based on computed tomography data.a-c, Cranial cartilages in estimated life position, with cartilages coloured in pairs in dorsal (a), ventral (b) and anterior (c) view.d,e, Mediolateral cartilages A in dorsal view (d) and median dorsal cartilage in ventral view (e) rendered with a vertical height map texture to emphasize the surface topology.f, Reconstruction of the forebrain relative to the cranial cartilages using a lamprey as a model 9,52 , shown in dorsal view.g, Cartilages in dorsal view, rendered transparent to show internal vasculature (red).h,i, Cartilages in preserved position in anterior view with dermal skeleton shown (h) and removed (i).Colours in a,b,c,f,h,i as in Fig. 1 with the following

Extended Data Fig. 1 |
Eriptychius PF 1795 shown in the context of the surrounding matrix.a,b, Tomograms showing sections through the part of the specimen preserved in the matrix.c-e, 3D render of the articulated Eriptychius specimen part preserved in epoxy relative to the matrix part, (c) in anatomical ventral view, (d) lateroventral view and (e) laterally with the matrix rendered transparent.Extended Data Fig. 2 | Photographs of Eriptychius PF 1795.a, part preserved in matrix contrasting our interpretation with that of Denison 8 .b, a close up of the front part of the specimen showing the vasculature.c, the part of the specimen preserved in epoxy with d, a close up of the 'rostral plates and e, a close up of the orbital plates.Roman case labels, our interpretation; italics case labels, Denison's interpretation corresponding to figure 2 of ref.8.Extended Data Fig. 3 | Eriptychius PF 1795 tissues in tomographic section.Section of tomogram from the higher resolution scan set in coronal plane showing dermal 'rostral' plates overlying and wrapping Extended Data Fig. 4 | 3D render of Eriptychius PF 1795.a, in ventral view.b, in dorsal view.c, chain of orbital plates in visceral view, showing curved visceral surfaces.d,e, the best-visualized orbital plate in (d) lateral and (e) visceral view.f, g, left branchial plate in (f) lateral and (g) anterior view.h,?R.branchial plate fragment in lateral view.i, right branchial plate in lateral view.Colour scheme and abbreviations as in Figs. 1 and 2, except: grey, cartilages; blue-green, dermal skeleton.Lighter shades denote material from PF 1795a, darker shades PF 1795b.Extended Data Fig. 5 | 3D renders of the individual endocranial elements of Eriptychius PF 1795.a, median dorsal cartilage in anterior view; b-d median ventral cartilage in (b) posterior, (c) left lateral and (d) dorsal view; e,f mediolateral cartilages A in (e) ventral and (f), posterior view; g,h both surfaces of mediolateral cartilages B; i,j both surfaces of mediolateral cartilages C; k-m, left orbital cartilage in (k) medial, (l) posterior and (m) latero-posterior view.Colour scheme and abbreviations as in Fig. 2. Extended Data Fig. 6 | Additional 3D renders of the vascularization of the endocranium of Eriptychius PF 1795.a,b, median dorsal cartilage in (a) dorsal and (b) ventral view; c,d, median ventral cartilage in (c) ventral and (d) dorsal view; e,f, mediolateral cartilages A in (e) dorsal and (f) ventral view; g, mediolateral cartilages B; h, mediolateral cartilages C; i,j left orbital cartilage in (i) dorsal and (j) ventral view; k,l right orbital cartilage in (k) dorsal and (l) ventral view.Colour scheme as in Fig. 2. Extended Data Fig. 7 | Strict consensus result of the parsimony analysis.Strict consensus tree resulting from the parsimony phylogenetic analysis described in the methods.Extended Data Fig. 8 | Adams consensus result of the parsimony analysis.Adams consensus tree resulting from the parsimony phylogenetic analysis described in the methods.Extended Data Fig. 9 | Consensus result of the Bayesian analysis.Majority rule consensus tree resulting from the Bayesian phylogenetic analysis described in the methods.Node values correspond to posterior probabilities.nature portfolio | reporting summary April 2023 Corresponding author(s): Richard P. Dearden Last updated by author(s): Aug 1, 2023Reporting SummaryNature Portfolio wishes to improve the reproducibility of the work that we publish.This form provides structure for consistency and transparency in reporting.For further information on Nature Portfolio policies, see our Editorial Policies and the Editorial Policy Checklist.