New evidence for mammaliaform ear evolution and feeding adaptation in a Jurassic ecosystem

Journal name:
Nature
Volume:
548,
Pages:
326–329
Date published:
DOI:
doi:10.1038/nature23483
Received
Accepted
Published online

Stem mammaliaforms are forerunners to modern mammals1, and they achieved considerable ecomorphological diversity in their own right2. Recent discoveries suggest that eleutherodontids, a subclade of Haramiyida, were more species-rich during the Jurassic period in Asia than previously recognized3, 4, 5, 6, 7, 8, 9, 10, 11, 12. Here we report a new Jurassic eleutherodontid mammaliaform with an unusual mosaic of highly specialized characteristics1, 2, 3, 4, 5, 6, and the results of phylogenetic analyses that support the hypothesis that haramiyidans are stem mammaliaforms. The new fossil shows fossilized skin membranes that are interpreted to be for gliding and a mandibular middle ear with a unique character combination previously unknown in mammaliaforms. Incisor replacement is prolonged until well after molars are fully erupted, a timing pattern unique to most other mammaliaforms. In situ molar occlusion and a functional analysis reveal a new mode of dental occlusion: dual mortar–pestle occlusion of opposing upper and lower molars, probably for dual crushing and grinding. This suggests that eleutherodontids are herbivorous, and probably specialized for granivory or feeding on soft plant tissues. The inferred dietary adaptation of eleutherodontid gliders represents a remarkable evolutionary convergence with herbivorous gliders in Theria. These Jurassic fossils represent volant, herbivorous stem mammaliaforms associated with pre-angiosperm plants that appear long before the later, iterative associations between angiosperm plants and volant herbivores in various therian clades.

At a glance

Figures

  1. Dentition of mammaliaform Vilevolodon diplomylos (Haramiyida, Eleutherodontidae) (BMNH2942 holotype).
    Figure 1: Dentition of mammaliaform Vilevolodon diplomylos (Haramiyida, Eleutherodontidae) (BMNH2942 holotype).

    a, b, Mandibles and teeth on BMNH2942A from computed tomography scans (top and under sides); lower M2 (inset) from BMNH2042B. c, Upper teeth in lingual and occlusal views. d, Lower teeth in lingual and occlusal views. e, Posteromedial view of the mandible to demonstrate the M2/M2 occlusal alignment and the middle ear contact to mandible. f, Dual mortar–pestle occlusion. The tallest cusp ‘A1 pestle’ of lower M2 occludes into the anterior fusiform mortar basin of M2 while the tallest cusp ‘A1 pestle’ of upper M2 occludes into the posterior fusiform mortar basin of lower M2. g, M2/M2 dual mortar–pestle occlusion in posteromedial view as in e. L, left; R, right; P, premolars; M, molars. Lower incisors I1A and I1B are two successive incisor generations of lower locus 1; incisors I1A and I1B are two generations at upper incisor 1 locus. The whole slab and counter-slab with carbonized patagia are shown in Extended Data Fig. 1.

  2. Disparate tooth occlusal patterns in haramiyidans.
    Figure 2: Disparate tooth occlusal patterns in haramiyidans.

    a, Haramiyavia with lower A1 of M3 occluding into lingual embrasure of upper M2–M3. Upper tooth outlines flipped and superimposed on lower molar M3 (ref. 6). b, Occlusal view of right M3. c, Occlusal view of right upper M2–M3. d, Megaconus. Lower M3 A1 cusp occludes into the lingual embrasure between upper M2–M3. e, f, Right M3 occlusal view (e) and right M1–M3 occlusal view (f). g, Maiopatagium right upper P3–M2 in occlusal view. h–j, Vilevolodon right lower M1–M2 (i) and right upper P4–M2 (j). The opposite upper molars (transparent) over the lower molars (grey) in dual mortar-pestle occlusion. km, Eleutherodon putative right M2 (l) and right M2 (m)3, 9. M2 (transparent) over M2 (grey)3, 4. Node 1, ancestral occlusal pattern of the haramiyidan Haramiyavia: thoroughfare occlusal furrow for ortho-palinal occlusal movement. Node 2, clade of Megaconus and Vilevolodon: shortened Meckel’s sulcus and reduced postdentary trough. Node 3, ventro-flexed rostrum. Node 4, cusp coalescence within cusp row, thoroughfare occlusal furrow for ortho-palinal occlusal movement. Node 5, clade of Vilevolodon and Eleutherodon (including Xianshou, Arboroharamiya and Sineleutherus) with dual mortar–pestle occlusion. Phylogenetic analyses presented in Supplementary Information.

  3. Transformations of mandibular and middle ear structures among mammaliaforms.
    Figure 3: Transformations of mandibular and middle ear structures among mammaliaforms.

    a, b, Multituberculate Sinobaatar mandible (a) and middle ear (b). The Meckel’s element and the anterior limb of ectotympanic are absent from the middle ear, and the mandible has no structure for middle ear connection (BMNH1145 and other specimens). This represents one of at least three evolutionarily independent ear–jaw disconnections in crown mammals. DMME, definitive mammalian middle ear. c, Haramiyavia6 mandible. d, e, Haramiyidan Megaconus mandible (d) and middle ear (e)9. fh, Vilevolodon middle ear removed from right mandible to show attachment structure (f); the middle ear restored as on the mandible (g); and the middle ear reconstruction showing Meckel’s element continuing with the malleus and the ectotympanic with an anterior limb (h). Details of fossil and comparison are provided in Extended Data Figs 2, 8 and 9. i, The cynodont Kayentatherium30, showing morphological similarities to Vilevolodon. jm, Mammaliaforms Morganucodon (j, k) Hadrocodium (l) and Agilodocodon (m), and their mandibular structure for middle ear attachment. The haramiyidan clade (ch) represents an independent evolution in the size reduction of Meckel’s element and shortening of the anterior ectotympanic limb, but these are still attached to a vestigial Meckel’s sulcus and a reduced postdentary trough. Vilevolodon resembles Kayentatherium (i) and other mammaliaforms but differs from multituberculates with regards to the middle ear (a, b, fh) (see also Extended Data Figs 2, 8 and 9).

  4. Mammaliaform Vilevolodon diplomylos (Haramiyida, Eleutherodontidae) holotype (Beijing Museum of Natural History PM002942).
    Extended Data Fig. 1: Mammaliaform Vilevolodon diplomylos (Haramiyida, Eleutherodontidae) holotype (Beijing Museum of Natural History PM002942).

    a, b, BMNH2942A (main slab). Skeletal feature identification, and the outline of the carbonized patagial skin membranes (indicated by red arrows). c, d, BMNH2942B (counterpart). The outline of the carbonized patagial membranes: propatagium, plagiopatagium and uropatagium (indicated by red arrows), and their anatomical relationship to skeleton. e, Partial cranial roof and facial bones preserved on BMNH2942B, extracted by computed tomography (CT) scans and 3D rendering. e1, M2 from counterpart BMNH2942B, in occlusal view. f, Right pes preserved on counterpart, with ventral view of tarsals and approximately lateral views of digit long bones (metatarsal 1 and digit 1 phalanges are not preserved).

  5. Vilevolodon skull on holotype main part (BMNH2942A).
    Extended Data Fig. 2: Vilevolodon skull on holotype main part (BMNH2942A).

    a, Stereo pair photographs of skull structures after preliminary preparation. be, CT scan rendering of BMNH2942A, viewed from the partially exposed top side of the skull. Dark green indicates bones segmented from counterpart BMNH2942B. b, Intact left zygoma. c, Stereo images of the skull with left zygoma removed to show the rostroventral flexion of maxillary and the premolar–molar row. d, Unexposed underside of the skull of BMN2942A visualized by CT segmentation, with right mandible in place. e, Stereo images of the skull with right mandible removed to show rostroventral flexion of right maxillary and its tooth row. ** indicates the irregular depression on the mandible formed by underlying massive tooth roots; but it is not a muscle fossa. Detailed analysis of the ear is provided in Extended Data Fig. 8, and comparative morphology in Extended Data Figs 8 and 9.

  6. Vilevolodon mandible and teeth on holotype main part (BMNH2942A).
    Extended Data Fig. 3: Vilevolodon mandible and teeth on holotype main part (BMNH2942A).

    Skull bones removed to expose the roots of upper teeth. a, CT scan rendering of BMNH2942A, viewed from the partially exposed top side of the skull; the intact occlusion of right upper and lower molars. Left dentary condyle and coronoid were preserved on the counterpart, helping to expose the full middle ear. b, CT rendering of the unexposed underside of the skull. c, Top side of the skull of BMNH2942A. The left mandible composite from the coronoid process and dentary condyle segmented from CT scans of the BMNH2942B counterpart. The left petrosal, the occipital and cervicals rendered invisible to highlight the relationship of middle ear to the mandible.

  7. Vilevolodon dentition.
    Extended Data Fig. 4: Vilevolodon dentition.

    a, Right upper teeth in lingual view. The dashed line shows the en echelon or step-wise pattern occlusal surfaces along the upper tooth row, a plesiomorphy of haramiyidans. b, Right upper teeth in ventral view (stereo pair photographs) with black arrows indicating successive lingual imbrication of M1 and M2. c, Left upper teeth (stereo pair photos). Note that M2 was compressed in fossilization. d, Left upper teeth in labial view (stereo pair photos). The dashed line indicates P3–P4 flexure. e, Exposed left lower teeth: first generation deciduous lower incisor 1 (di-1a), second generation deciduous incisor 1 (i-1b), ultimate premolar P4 and M1 extracted from BMNH2942A. The M2 was extracted from BMNH2942B. f, Left lower teeth (stereo pair photos) in occlusal view. Note the roots of deciduous incisors are entirely medial to roots of premolar or molars. The M2 was extracted from BMNH2942B, and all other teeth are from BMNH2942A. g, Right lower M2 from BMNH2942B. h, Right lower M2 (stereo images, occlusal view). i, Right lower teeth (stereo images) in occlusal view from BMNH2942A. j, Right lower teeth in medial view. Bones removed digitally to expose tooth roots and the unerupted replacing incisors, highlighting the prolonged replacement of incisors relative to the adult premolars and molars.

  8. Dual mortar–pestle occlusion of Vilevolodon molars, in contrast to embrasure occlusion of Haramiyavia.
    Extended Data Fig. 5: Dual mortar–pestle occlusion of Vilevolodon molars, in contrast to embrasure occlusion of Haramiyavia.

    a, Vilevolodon right mandible in oblique posterior view, highlighting the M2/M2 orientation and the attachment of the mandibular middle ear. b, Cusp pattern of upper molars (in translucent outline). Note that M2 is imbricated more lingually to M1. c, Cusp pattern of lower molars. Note that M2 is imbricated more lingually to M1, mirroring the imbrication of upper M2 to M1. d, Superimposition of upper M1–M2 (translucent outline) over lower M1–M2 (grey) for dual mortar–pestle occlusion of the upper and lower molars (see also Fig. 3). e, Oblique posterior view of right M2 and M2 in dual mortar–pestle occlusion: A1 cusp (A1 pestle) entering the M2 basin and the A1 cusp (A1 pestle) entering the M2 basin. f, Dual mortar–pestle occlusion of left M2 and M2 (solid surface models) in occlusal views. g, Dual mortar–pestle occlusion of right M2 and M2 surface models in medial (lingual) view (g1), posterior (distal) view (g2), lateral (labial) view (g3), and anterior (mesial) view (g4). h, Occlusion of right lower M1–M2 to right upper M1–M2 in lingual view, based on the best fit of opposing teeth. Note that the B4–B3 cusps of upper molars are always lingual to the lower lingual cusp row, but the B1 cusp is always labial to lower lingual cusp row and in the lower median furrow. i, Occlusion of right M1–M2 to right M1–M2 in labial view. Note that cusp A4 of the upper molar is always labial (lateral) to cusp A1 of the lower molar. The upper labial row A1–A4 occludes obliquely over lower labial row B1–B4. There is no longitudinal alignment of cusp rows between the opposite upper and lower teeth, as in Thomasia, Haramiyavia and Maiopatagium (Fig. 3). j, Haramiyidan Haramiyavia occlusal trajectory (blue, orthal occlusion; green, palinal movement). k, Haramiyavia: transition of orthal phase (blue) to palinal phase (green) occlusion. Cusp row A1–A4 of the lower molar is lingual to cusp row B1–B5 of the upper molars. l, Haramiyavia: full orthal occlusion of M3 with A1 cusp in embrasure between upper M2–M3, and cusp row A1–A4 lingual to cusps B1–B5.

  9. Autapomorphic occlusal features of Vilevolodon, including two inferred cycles of the molar occlusal movement.
    Extended Data Fig. 6: Autapomorphic occlusal features of Vilevolodon, including two inferred cycles of the molar occlusal movement.

    The upper tooth row has four teeth and is longer than the lower tooth row of three teeth. The upper P3–P4 flexure forms a prominent angle of the occlusal planes between the two premolars. In cycle 1 when lower M1–M2 is able to make full contact with M1–M2, P4 and P3 have no contact. In cycle 2, P4/P3 can make full contact, and their occlusion occurs only during cycle 2. But during this cycle, when the primary cusp of the P4 makes full occlusal contact with P3, the lower M1 and M2 can only barely contact the upper M1 and M2 and cannot make full contact. Occlusal movement was simulated by animation of STL models (see Supplementary Video 1) and inferred by maximal fit of upper and lower tooth occlusal surfaces. Blue arrows indicate the path of cusp A1 of the lower premolar. The blue band in the central column shows the differences in the position of maximum intercuspation (centric occlusion of the P4 primary cusp) between cycle 1 (above) and cycle 2 (below).

  10. Dental and mandibular morphologies of eleutherodontids and extant frugivorous bats Sturnira and Artibeus (Chiroptera, Phyllostomidae), Hypsignathus (Megachiroptera: Pteropodidae), and frugivorous/omnivorous primate Daubentonia.
    Extended Data Fig. 7: Dental and mandibular morphologies of eleutherodontids and extant frugivorous bats Sturnira and Artibeus (Chiroptera, Phyllostomidae), Hypsignathus (Megachiroptera: Pteropodidae), and frugivorous/omnivorous primate Daubentonia.

    ac, Vilevolodon left M2 (a) and left P4 (b, c). df, Lower teeth of frugivorous bats: Sturnira lilium lilium (FMNH105870) (d); Artibeus jamaicensis (FMNH30776) (e); and Hypsignathus monstrosus (University of Chicago Biological Sciences Division Teaching Collection) (f). g, Megaconus right mandible in lateral view, highlighting the zigzag tooth row profile9. h, Vilevolodon right mandible in lateral view, highlighting the zigzag tooth row profile. i, Hypsignathus mandible in lateral view. j, Artibeus mandible in lateral view, highlighting the zigzag tooth row profile. k, Primate Daubentonia (FMNH15529) right mandible in lateral view. a, df, Similarities of lower molars of Vilevolodon with the basined lower molar talonids of frugivorous bats Artibeus, Sturnira and Hypsignathus. Eleutherodontids are particularly similar to the highly creased talonid basins of Artibeus, whereas Maiopatagium is more similar to Sturnira and pteropodid bats in molar pattern. b–e, f, Similarities of lower premolar of Vilevolodon to those of Sturnira and Hypsignathus. g, h, Similarity of the zigzag tooth row profile of lower premolars and molars of Megaconus and eleutherodontids to those of frugivorous bats. g, h, k, Similarity in mandibles of eleutherodonts and the primate Daubentonia (FMNH15529), which is frugivorous and insectivorous.

  11. Vilevolodon middle ear structure preserved in holotype (BMNH2942) and interpretive reconstruction.
    Extended Data Fig. 8: Vilevolodon middle ear structure preserved in holotype (BMNH2942) and interpretive reconstruction.

    a, BMNH2942A in approximately ventral view. External and exposed aspect of left petrosal (partial) and middle ear bones as preserved. The dentary and cervical vertebrae are digitally removed so that the view of the middle ear features is unobstructed. b, BMNH2942A ‘internal’ view (underside of the fossil in matrix), showing the petrosal (partially segmented) and middle ear bones. c, Interpretative reconstruction of the middle ear bones of Vilevolodon, based on middle ear bones exposed in CT scans and 3D segmentation. d, A composite reconstruction of the middle ear of eutriconodonts. e, Cynodont Kayentatherium (redrawn from ref. 30). f, Pre-mammaliaform cynodont Cynognathus middle ear (redrawn from ref. 31) Note 1 to readers: the gracile ectotympanic (reflected lamina of angular), and this gracile and relatively straight morphology is similar to the reflected lamina of several non-mammalian cynodonts. The angle of attachment to the anterior–posterior limbs of ectotympanic (homologue of the angular) is slightly distorted by fossilization. Notes 2 and 3 to readers: the manubrial portion of the malleus (Note 2) is partly attached to the malleus body (Note 3), but there is a gap between these two structures owing to imperfect preservation. Note 4: the posterior end of Meckel’s cartilage is attached to a large sliver of bone. The latter is tentatively interpreted to be a surangular, which in non-mammalian cynodonts is typically parallel to the gonial part (the prearticular) of the malleus (although separated from the latter). There appears to be a suture between the prearticular part of the malleus, and the putative surangular. We infer that the Meckel’s cartilage and the ectotympanic, as preserved in close association, contact each other loosely. But they became slightly separated from each other during fossilization. We reconstructed these two elements as contiguous, as in cynodonts and other mammaliaforms, as demonstrated in c.

  12. Comparative morphology of the mandibles and partial mammalian middle ears of extinct Mesozoic mammaliaforms.
    Extended Data Fig. 9: Comparative morphology of the mandibles and partial mammalian middle ears of extinct Mesozoic mammaliaforms.

    a, Morganucodon and several mammaliaforms possess plesiomorphic mandibular middle ear of cynodonts (MMEC). The ectotympanic (homologue of the angular) is nestled in the angular concavity of the mandible. The surangular and the gonial part (homologue of the prearticular) of the malleus make full contact with a broad postdentary trough that extends posteriorly to near the dentary condyle under the medial ridge. The Meckel’s element (ossified cartilage) starts from the postdentary trough, passes anteriorly below the mandibular foramen, and extends further anteriorly into a long Meckel’s sulcus below the mandibular tooth row. b, Eleutherodontid Vilevolodon shows a mandibular connection of the middle ear, although much more reduced (antero-posteriorly shortened) than those of other mammaliaforms. The anterior limb of the ectotympanic and the gonial part (prearticular) of the malleus are nestled in a reduced postdentary trough on the internal aspect of the inflected mandibular angle. The groove occupies the identical position of the postdentary trough in the angular region of other mammaliaforms. However, it is narrower and much shorter than the postdentary trough of other mammaliaforms. The trough is shorter (an apomorphic trait) and does not extend posteriorly to the dentary condyle, in contrast to the postdentary trough in other mammaliaforms. The short and tapering Meckel’s cartilage ends below the mandibular foramen, but it does not extend further anteriorly to the mandibular body below the tooth row. The Meckel’s element and its corresponding Meckel’s sulcus are much shorter than those of other mammaliaforms, eutriconodonts and spalacotherioids. The gonial portion of the malleus (and the putative surangular) also contacts the ventral margin of the mandible. We infer that, in Vilevolodon, the ectotympanic (retroarticular process) and the manubrial part of the malleus are rotated medially and horizontally, away from the mandible (b3), as in eutriconodonts. c, Eutriconodont Yanoconodon, highlighting the massive ossified Meckel’s cartilage that connects the middle ear to the mandible. Owing to the curvature and mid-length bending of Meckel’s element, the ectotympanic and malleal manubrium rotated medially and away from the mandible32.

Videos

  1. Video 1: Analysis of tooth occlusion of Vilevolodon by animation of STL tooth models from CT scan
    Video 1: Video 1: Analysis of tooth occlusion of Vilevolodon by animation of STL tooth models from CT scan
    Part 1: Comparison of distinctive cycle 1 versus cycle 2 of premolars and molars: cycle 1 has full contacts of molars and incisors, but no contact between P3/p4; cycle 2 has full contact of P3/p4 but minimal or no contacts of molars and incisors. Part 2: Molar cusp and basin identification, and cycle 1 of occlusal movement of M1/m1 and M2/m2 during (see Fig. S6). During occlusal cycle 2, the P3/p4 are the only teeth that can fully occlude; the dual mortar-pestle contacts of upper and lower molars do not occur. Thus, cycle 2 is not shown here in animation analysis Part 2 of the video. For better clarity, the un-erupted deciduous incisors are not shown in the video.

Main

Clade Mammaliaformes1

Clade Haramiyida (emended by ref. 11)

Family Eleutherodontidae4, 11

Vilevolodon diplomylos gen. et sp. nov.

Etymology. Vilevol (Latin): glider; don (Greek): tooth, and a common suffix for mammalian taxon names; diplo (Greek): double; mylos (Greek): grinding; diplomylos refers to the dual mortar–pestle occlusion of opposing upper and lower molars.

Holotype. Beijing Museum of Natural History PM002942A and B (abbreviated as BMNH2942A and BMNH2942B) are the main slab and counter slab, respectively, preserved with carbonized residues of patagial skin membranes associated with a postcranial skeleton that is 70% complete. The skull is preserved with teeth in occlusion and a middle ear associated with the mandible (Figs 1, 2, Extended Data Figs 1, 2, 3).

Figure 1: Dentition of mammaliaform Vilevolodon diplomylos (Haramiyida, Eleutherodontidae) (BMNH2942 holotype).
Dentition of mammaliaform Vilevolodon diplomylos (Haramiyida, Eleutherodontidae) (BMNH2942 holotype).

a, b, Mandibles and teeth on BMNH2942A from computed tomography scans (top and under sides); lower M2 (inset) from BMNH2042B. c, Upper teeth in lingual and occlusal views. d, Lower teeth in lingual and occlusal views. e, Posteromedial view of the mandible to demonstrate the M2/M2 occlusal alignment and the middle ear contact to mandible. f, Dual mortar–pestle occlusion. The tallest cusp ‘A1 pestle’ of lower M2 occludes into the anterior fusiform mortar basin of M2 while the tallest cusp ‘A1 pestle’ of upper M2 occludes into the posterior fusiform mortar basin of lower M2. g, M2/M2 dual mortar–pestle occlusion in posteromedial view as in e. L, left; R, right; P, premolars; M, molars. Lower incisors I1A and I1B are two successive incisor generations of lower locus 1; incisors I1A and I1B are two generations at upper incisor 1 locus. The whole slab and counter-slab with carbonized patagia are shown in Extended Data Fig. 1.

Figure 2: Disparate tooth occlusal patterns in haramiyidans.
Disparate tooth occlusal patterns in haramiyidans.

a, Haramiyavia with lower A1 of M3 occluding into lingual embrasure of upper M2–M3. Upper tooth outlines flipped and superimposed on lower molar M3 (ref. 6). b, Occlusal view of right M3. c, Occlusal view of right upper M2–M3. d, Megaconus. Lower M3 A1 cusp occludes into the lingual embrasure between upper M2–M3. e, f, Right M3 occlusal view (e) and right M1–M3 occlusal view (f). g, Maiopatagium right upper P3–M2 in occlusal view. h–j, Vilevolodon right lower M1–M2 (i) and right upper P4–M2 (j). The opposite upper molars (transparent) over the lower molars (grey) in dual mortar-pestle occlusion. km, Eleutherodon putative right M2 (l) and right M2 (m)3, 9. M2 (transparent) over M2 (grey)3, 4. Node 1, ancestral occlusal pattern of the haramiyidan Haramiyavia: thoroughfare occlusal furrow for ortho-palinal occlusal movement. Node 2, clade of Megaconus and Vilevolodon: shortened Meckel’s sulcus and reduced postdentary trough. Node 3, ventro-flexed rostrum. Node 4, cusp coalescence within cusp row, thoroughfare occlusal furrow for ortho-palinal occlusal movement. Node 5, clade of Vilevolodon and Eleutherodon (including Xianshou, Arboroharamiya and Sineleutherus) with dual mortar–pestle occlusion. Phylogenetic analyses presented in Supplementary Information.

Geological age. Vilevolodon diplomylos type specimen is from the Nanshimen site of the Tiaojishan Formation in Qinglong County, Hebei Province, China, stratigraphically correlated by the regional index fossil Qaidamestheria13, and estimated to be 161–160 million years old. The Tiaojishan Formation has yielded additional mammaliaforms at the same site and elsewhere9, 10, 11, 14, 15, 16.

Diagnosis. Among haramiyidans, Vilevolodon is most similar to Arboroharamiya10 and Xianshou11 in having dual mortar–pestle molar occlusion in which the tallest distal cusp A1 of the upper molar (upper ‘pestle’) occludes into the distal basin (lower ‘mortar’) of the opposite lower molar. The tallest mesial cusp A1 of the lower molar (lower pestle) occludes into the mesial basin (upper mortar) of the opposite upper molar (Figs 1, 2, Extended Data Figs 3, 4, 5, 6; Supplementary Video 1). It differs from Maiopatagium15 and Shenshou11 in that upper molars are arranged in successively imbricated series in which M2 is oblique to M1, and M1 is oblique to P4. This is mirrored by imbrication of the lower teeth in which M2 is medial (and oblique) to M1, and M1 is medial to P4 (Fig. 1). This differs from the straight tooth rows of Maiopatagium and Shenshou (Fig. 2). Among eleutherodontids (see Supplementary Information), V. diplomylos is most similar to species of Xianshou in having a hypertrophied P4 and pronounced flexure of upper P3–P4 that is correlated with rostroventral bending of the maxilla. Differs from Xianshou species in having larger teeth but a shorter mandible. Differs from paulchoffatiid multituberculates16 in that its molars have a confluent root in a single alveolus, not double roots in separate alveoli. Differs from eutriconodonts and spalacotherioids of crown mammals in that Meckel’s sulcus is vestigial and Meckel’s element is shortened17, 18. Full differential diagnosis of V. diplomylos from other mammaliaforms is provided in the Supplementary Information.

The unique dental occlusion of Vilevolodon includes dual crushing and grinding functions. Our simulation analysis using STL models from computed tomography scans shows evidence of complex chewing movement in two separate occlusal cycles (Supplementary Video 1 and Extended Data Fig. 6). The trenchant cusp of P4 contacts the P4 basin simultaneously with the dual mortar–pestle occlusion of M1/M1 and M2/M2 in the same chewing cycle (Supplementary Video 1 and Extended Data Fig. 6). The main trajectory of cusp A1 (pestle) of the lower molar is ortho-palinal, similar to that of Haramiyavia4, 6, but the palinal movement has a strong ventral vector, as constrained by the deep basins of upper and lower molars. It is not possible for the lower P4–M2 to move posteriorly and horizontally (that is, fully palinal). This differs from multituberculates, Megaconus and Maiopatagium, in which the lower teeth can have full palinal movement4, 9, 19 (Fig. 2). The dual mortar–pestle occlusal contact and two distinctive occlusal cycles are unique to eleutherodontids (probably also applicable to Arboroharamiya), but absent in Haramiyavia, Megaconus, Maiopatagium and other stem mammaliaforms.

We infer that Vilevolodon had an herbivorous or omnivorous diet consisting largely of seeds and soft plant parts. Tooth crown complexity of Vilevolodon and Xianshou is comparable to derived multituberculates with plant-dominated diets20. Their tooth crenulations and creases resemble those of sciurid rodents that have diets of nuts, seeds, fruits and young leaves, supplemented by insects21. Vilevolodon was not a folivore, as it lacks strong tooth crests that are characteristic of primarily folivorous and volant taxa such as dermopterans, anomalurids and marsupial gliders. All modern mammalian gliders are primarily herbivorous and none are primarily insectivorous22, and we interpret Vilevolodon to be a glider based on the presence of carbonized patagia and skeletal morphometric analyses15. Thus, our herbivorous dietary inference for Vilevolodon is consistent with modern, analogous gliders.

The peculiar teeth of Vilevolodon expand the known dental morphological disparity of mammaliaforms. Its densely ornamented and partly basined teeth are markedly distinct from the simplified teeth of Maiopatagium and from the straight cusp rows and straight furrow of Megaconus. Jurassic eleutherodontids are highly transformed, even compared to Triassic haramiyids of the same clade (Fig. 2). Their disparate tooth morphologies suggest resource partitioning among omnivorous and herbivorous feeding guilds of eleutherodonts. Furthermore, eleutherodonts diversified contemporaneously with early multituberculates16 and ecomorphologically diverse docodonts14, 23, offering compelling evidence that clade divergence and ecological diversification are coupled in adaptive diversifications in multiple Mesozoic mammaliaform clades9, 10, 11 (Fig. 2).

Eleutherodonts diversified in a pre-angiosperm biota of the Middle to Late Jurassic. Their diets would probably have included seeds, reproductive parts such as strobili and cones of ferns, plus soft meristem tissues and young leaves of seed and gymnosperm plants, which have all been hypothesized to be probable dietary sources for animals24. The association of herbivorous and volant eleutherodonts with pre-angiosperm plants in the Jurassic is analogous to the association of herbivorous volant therians (that is, dermopterans, and multiple clades of rodents and marsupials) with angiosperm plants of the Cenozoic era. Eleutherodontids show a marked similarity to the primate Daubentonia in the ventrally bent rostrum and deep mandible, and both features are interpreted to be reinforcement for incisor gnawing25 (Extended Data Fig. 7). Daubentonia has a dietary mixture of fruits, exudates and insect larvae. Eleutherodonts and Megaconus (Extended Data Fig. 6) show a zigzag tooth row profile in lateral view, convergent with that of frugivorous bats26 (Extended Data Fig. 7). The highly ornamented teeth of eleutherodontids (Eleutherodon, Sineleutherus, Arboroharamiya, Xianshou and Vilevolodon) are convergent with some extant sciurid rodents with granivorous and frugivorous diets and to herbivorous phyllostomid bats21, 26. These observations suggest that haramiyidans differentiated into several feeding guilds during their evolutionary diversification in the Jurassic.

Computed tomography scans of the Vilevolodon holotype confirm that premolar loci have no replacing tooth. The right M2/M2 are fully erupted and occluded, and these have closed root tips. M1/M1 show occlusal wear (Extended Data Fig. 4). Cheek teeth of the Vilevolodon have attained adult status27 according to the individual dental age system detailed in ref. 28. However, its upper and lower incisors are still undergoing replacement, either by a prolonged, or a delayed replacement. Thus, Vilevolodon shows a heterochronical pattern of incisor replacement versus premolar and molar eruption, which is unique in mammaliaforms (except Sinoconodon)27. Either Vilevolodon had an unusually accelerated completion of molar eruptions as a juvenile (indicated by ongoing incisor replacement) or the ongoing incisor replacements at I1/I1 loci are a paedomorphic adult feature (indicated by complete molar eruption). By contrast, Morganucodon and docodontans have modern-mammal-like diphyodont replacements in which replacement of antemolars is completed during juvenile growth stages, well before complete eruption of adult molars23, 29. This suggests that diphyodont tooth replacement, a hallmark feature for determinate skull growth of modern mammals27, 29, had heterochronical variation, and may be homoplastic in early mammaliaform lineages.

In Vilevolodon the malleus is anteriorly connected to a short ossified Meckel’s cartilage (as in refs 17, 18) (which could be interpreted to be the prearticular bone; see Supplementary Information). The ectotympanic (homologous to the angular bone of non-mammalian cynodonts) has an anterior limb and a straight reflected lamina (Fig. 3 and Extended Data Figs 8 and 9). These ear structures are notably similar to those of tritylodontid cynodonts30. The anterior limb of the ectotympanic and short Meckel’s element are nestled in a triangular depression between the inflected angular process and vertical plate of the mandible. The homologous part of the postdentary trough in the mandibular angle in eleutherodonts represents only a small, reduced part of the full postdentary trough in Haramiyavia and other mammaliaforms (Fig. 3 and Extended Data Fig. 9). We confirm that eleutherodonts lack the anterior section of Meckel’s sulcus along the ventral margin of the mandible (beyond the mandibular foramen), as noted in earlier studies11, 12. Meckel’s cartilage is much smaller in eleutherodonts than in eutriconodonts and spalacotherioids. We infer that the middle ear would be nestled in the inflected mandibular angle (Extended Data Fig. 9). Vilevolodon differs from multituberculates in that multituberculates completely lost the anterior limb of the ectotympanic and the Meckel’s element (Fig. 3).

Figure 3: Transformations of mandibular and middle ear structures among mammaliaforms.
Transformations of mandibular and middle ear structures among mammaliaforms.

a, b, Multituberculate Sinobaatar mandible (a) and middle ear (b). The Meckel’s element and the anterior limb of ectotympanic are absent from the middle ear, and the mandible has no structure for middle ear connection (BMNH1145 and other specimens). This represents one of at least three evolutionarily independent ear–jaw disconnections in crown mammals. DMME, definitive mammalian middle ear. c, Haramiyavia6 mandible. d, e, Haramiyidan Megaconus mandible (d) and middle ear (e)9. fh, Vilevolodon middle ear removed from right mandible to show attachment structure (f); the middle ear restored as on the mandible (g); and the middle ear reconstruction showing Meckel’s element continuing with the malleus and the ectotympanic with an anterior limb (h). Details of fossil and comparison are provided in Extended Data Figs 2, 8 and 9. i, The cynodont Kayentatherium30, showing morphological similarities to Vilevolodon. jm, Mammaliaforms Morganucodon (j, k) Hadrocodium (l) and Agilodocodon (m), and their mandibular structure for middle ear attachment. The haramiyidan clade (ch) represents an independent evolution in the size reduction of Meckel’s element and shortening of the anterior ectotympanic limb, but these are still attached to a vestigial Meckel’s sulcus and a reduced postdentary trough. Vilevolodon resembles Kayentatherium (i) and other mammaliaforms but differs from multituberculates with regards to the middle ear (a, b, fh) (see also Extended Data Figs 2, 8 and 9).

In the phylogenetic context that haramiyidans (including eleutherodontids) are a clade of stem mammaliaforms that exclude multituberculates1, 5, 6, 15, the shortened Meckel’s element and reduced postdentary trough are separately derived features among haramiyidans, as the Late Triassic Haramiyavia and other mammaliaforms have a full Meckel’s sulcus and postdentary trough (Fig. 3). Size decrease of the middle ear among haramiyidans is convergent with the multiple evolutions of size reduction in crown mammal clades, except that the ear elements had not achieved full separation from the mandible in a dead-end side-branch lineage of mammaliaforms.

The unique mosaic of characters related to tooth replacements and the middle ear of eleutherodonts adds to growing evidence of complex transformations of mammalian characteristics. Their complex dentitions and occlusal patterns are probably adapted for omnivory and herbivory, showing that the volant and herbivorous lifestyle, previously known only in therian gliders, was also part of mammaliaform evolutionary experimentation during the Jurassic (Fig. 2 and Extended Data Fig. 7).

Data Availability

All specimens of this study have been deposited at the Beijing Museum of Natural History. Graphics and phylogenetics data are provided in the Supplementary Information. Life Science Identifier (LSID): the new genus and species are registered with Zoobank (http://zoobank.org): urn:lsid:zoobank.org:pub:10917E53-A185-48A5-8C53-A30C5D7474A5.

References

  1. Rowe, T. B. Definition, diagnosis, and origin of Mammalia. J. Vertebr. Paleontol. 8, 241264 (1988)
  2. Luo, Z.-X. Transformation and diversification in early mammal evolution. Nature 450, 10111019 (2007)
  3. Kermack, K. A. et al. New multituberculate-like teeth from the Middle Jurassic of England. Acta Palaeontol. Pol. 43, 581606 (1998)
  4. Butler, P. M. Review of the early allotherian mammals. Acta Palaeontol. Pol. 45, 317342 (2000)
  5. Jenkins, F. A. Jr, Gatesy, S. M., Shubin, N. H. & Amaral, W. W. Haramiyids and Triassic mammalian evolution. Nature 385, 715718 (1997)
  6. Luo, Z.-X., Gatesy, S. M., Jenkins, F. A. Jr, Amaral, W. W. & Shubin, N. H. Mandibular and dental characteristics of Late Triassic mammaliaform Haramiyavia and their ramifications for basal mammal evolution. Proc. Natl Acad. Sci. USA 112, E7101E7109 (2015)
  7. Martin, T. et al. Mammals from the Late Jurassic Qigu Formation in the southern Junggar Basin, Xinjiang, northwest China. Palaeodiv. et Palaeoenviron. 90, 295319 (2010)
  8. Averianov, A. O., Lopatin, A. V. & Krasnolutskii, S. A. The first Haramiyid (Mammalia, Allotheria) from the Jurassic of Russia. Dokl. Biol. Sci. 437, 103106 (2011)
  9. Zhou, C.-F., Wu, S., Martin, T. & Luo, Z.-X. A Jurassic mammaliaform and the earliest mammalian evolutionary adaptations. Nature 500, 163167 (2013)
  10. Zheng, X., Bi, S., Wang, X. & Meng, J. A new arboreal haramiyid shows the diversity of crown mammals in the Jurassic period. Nature 500, 199202 (2013)
  11. Bi, S., Wang, Y., Guan, J., Sheng, X. & Meng, J. Three new Jurassic euharamiyidan species reinforce early divergence of mammals. Nature 514, 579584 (2014)
  12. Meng, J., Bi, S., Zheng, X. & Wang, X. Ear ossicle morphology of the Jurassic euharamiyidan Arboroharamiya and evolution of mammalian middle ear. J. Morphol. (2016)
  13. Liao, H.-Y. et al. Micro-ornamentations on carapaces of Euestheria hingyuanensis (Crustaceaa: Spinificance) and its biostratigraphic significance. Acta Palaeontologica Sin. 53, 201216 (2014)
  14. Luo, Z.-X. et al. Mammalian evolution. Evolutionary development in basal mammaliaforms as revealed by a docodontan. Science 347, 760764 (2015)
  15. Meng Q.-J. et al. New gliding mammaliaforms from the Jurassic. Nature http://dx.doi.org/10.1038/nature23476 (2017)
  16. Yuan, C.-X., Ji, Q., Meng, Q.-J., Tabrum, A. R. & Luo, Z.-X. Earliest evolution of multituberculate mammals revealed by a new Jurassic fossil. Science 341, 779783 (2013)
  17. Luo, Z.–X. Developmental patterns in Mesozoic evolution of mammal ears. Annu. Rev. Ecol. Evol. Syst. 42, 355380 (2011)
  18. Meng, J., Wang, Y. & Li, C. Transitional mammalian middle ear from a new Cretaceous Jehol eutriconodont. Nature 472, 181185 (2011)
  19. Lazzari, V. et al. Occlusal pattern in paulchoffatiid multituberculates and the evolution of cusp morphology in mammaliamorphs with rodent-like dentitions. J. Mamm. Evol. 17, 177192 (2010)
  20. Wilson, G. P. et al. Adaptive radiation of multituberculate mammals before the extinction of dinosaurs. Nature 483, 457460 (2012)
  21. Thorington R. W. Jr et al. The difficulties of identifying flying squirrels (Sciuridae: Pteromyini) in the fossil record. J. Vertebr. Paleontol. 25, 950961 (2005)
  22. Jackson, S. M. & Schouten, P. Gliding Mammals of the World (CSIRO Publishing, 2012)
  23. Meng, Q.-J. et al. Mammalian evolution. An arboreal docodont from the Jurassic and mammaliaform ecological diversification. Science 347, 764768 (2015)
  24. Labandeira, C. C. The pollination of mid Mesozoic seed plants and the early history of long-proboscid insects. Ann. Mo. Bot. Gard. 97, 469513 (2010)
  25. Radinsky, L. B. A new approach to mammalian cranial analysis, illustrated by examples of prosimian primates. J. Morphol. 124, 167180 (1968)
  26. Santana, S. E. et al. The better to eat you with: functional correlates of tooth structure in bats. Funct. Ecol. 25, 839847 (2011)
  27. Luo, Z.-X., et al. Evolution of dental replacement in mammals. Carnegie Mus. Nat. Hist. Bull. 36, 159175 (2004)
  28. Anders, U. et al. Generalized individual dental age stages for fossil and extant placental mammals. Paläontol. Zeitschr. 85, 321339 (2011)
  29. O’Meara, R. N. & Asher, R. J. The evolution of growth patterns in mammalian versus nonmammalian cynodonts. Paleobiology 42, 439464 (2016)
  30. Sues, H.-D. The skull and dentition of two tritylodontid synapsids from the Lower Jurassic of western North America. Bull. Mus. Comp. Zool. 151, 217268 (1986)
  31. Kermack, K. A., Mussett, F. & Rigney, H. W. The lower jaw of Morganucodon. Zool. J. Linn. Soc. 53, 87175 (1973)
  32. Luo, Z.-X., Chen, P., Li, G. & Chen, M. A new eutriconodont mammal and evolutionary development in early mammals. Nature 446, 288293 (2007)

Download references

Acknowledgements

We thank A. Shinya for fossil preparation; S. Bi, S. Gatesy, L. Heaney, H.-J. Li, Z.-J. Gao, T. Martin, B. Patterson, N. Shubin, X.-T. Zheng and C.-F. Zhou for access to comparative specimens. Funding supported Q.-J.M. (Beijing Scientific Commission) and Z.-X.L. (UChicago-BSD). Full acknowledgments are provided in the Supplementary Information.

Author information

Affiliations

  1. Department of Organismal Biology and Anatomy, The University of Chicago, Chicago, Illinois 60637, USA

    • Zhe-Xi Luo &
    • April I. Neander
  2. Committee on Evolutionary Biology, The University of Chicago, Chicago, Illinois 60637, USA

    • Zhe-Xi Luo &
    • David M. Grossnickle
  3. Beijing Museum of Natural History, Beijing 100050, China

    • Qing-Jin Meng,
    • Di Liu &
    • Yu-Guang Zhang
  4. Hebei GEO University, Shijiazhuang 050031, Hebei Province, China

    • Qiang Ji

Contributions

Q.-J.M. and Z.-X.L. conceived the project; Q.-J.M., Y.-G.Z., D.L. and Q.J. acquired fossils and studied stratigraphy; all authors were involved in lab fossil work and interpretation; Z.-X.L. and D.M.G. did phylogenetic analyses; A.I.N. scanned and prepared graphics of fossils; Z.-X.L. composed figures; Z.-X.L., Q.-J.M. and D.M.G. led the writing, with feedback from all authors.

Competing financial interests

The authors declare no competing financial interests.

Corresponding authors

Correspondence to:

Reviewer Information Nature thanks G. Rougier and the other anonymous reviewer(s) for their contribution to the peer review of this work.

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Author details

Extended data figures and tables

Extended Data Figures

  1. Extended Data Figure 1: Mammaliaform Vilevolodon diplomylos (Haramiyida, Eleutherodontidae) holotype (Beijing Museum of Natural History PM002942). (547 KB)

    a, b, BMNH2942A (main slab). Skeletal feature identification, and the outline of the carbonized patagial skin membranes (indicated by red arrows). c, d, BMNH2942B (counterpart). The outline of the carbonized patagial membranes: propatagium, plagiopatagium and uropatagium (indicated by red arrows), and their anatomical relationship to skeleton. e, Partial cranial roof and facial bones preserved on BMNH2942B, extracted by computed tomography (CT) scans and 3D rendering. e1, M2 from counterpart BMNH2942B, in occlusal view. f, Right pes preserved on counterpart, with ventral view of tarsals and approximately lateral views of digit long bones (metatarsal 1 and digit 1 phalanges are not preserved).

  2. Extended Data Figure 2: Vilevolodon skull on holotype main part (BMNH2942A). (828 KB)

    a, Stereo pair photographs of skull structures after preliminary preparation. be, CT scan rendering of BMNH2942A, viewed from the partially exposed top side of the skull. Dark green indicates bones segmented from counterpart BMNH2942B. b, Intact left zygoma. c, Stereo images of the skull with left zygoma removed to show the rostroventral flexion of maxillary and the premolar–molar row. d, Unexposed underside of the skull of BMN2942A visualized by CT segmentation, with right mandible in place. e, Stereo images of the skull with right mandible removed to show rostroventral flexion of right maxillary and its tooth row. ** indicates the irregular depression on the mandible formed by underlying massive tooth roots; but it is not a muscle fossa. Detailed analysis of the ear is provided in Extended Data Fig. 8, and comparative morphology in Extended Data Figs 8 and 9.

  3. Extended Data Figure 3: Vilevolodon mandible and teeth on holotype main part (BMNH2942A). (498 KB)

    Skull bones removed to expose the roots of upper teeth. a, CT scan rendering of BMNH2942A, viewed from the partially exposed top side of the skull; the intact occlusion of right upper and lower molars. Left dentary condyle and coronoid were preserved on the counterpart, helping to expose the full middle ear. b, CT rendering of the unexposed underside of the skull. c, Top side of the skull of BMNH2942A. The left mandible composite from the coronoid process and dentary condyle segmented from CT scans of the BMNH2942B counterpart. The left petrosal, the occipital and cervicals rendered invisible to highlight the relationship of middle ear to the mandible.

  4. Extended Data Figure 4: Vilevolodon dentition. (485 KB)

    a, Right upper teeth in lingual view. The dashed line shows the en echelon or step-wise pattern occlusal surfaces along the upper tooth row, a plesiomorphy of haramiyidans. b, Right upper teeth in ventral view (stereo pair photographs) with black arrows indicating successive lingual imbrication of M1 and M2. c, Left upper teeth (stereo pair photos). Note that M2 was compressed in fossilization. d, Left upper teeth in labial view (stereo pair photos). The dashed line indicates P3–P4 flexure. e, Exposed left lower teeth: first generation deciduous lower incisor 1 (di-1a), second generation deciduous incisor 1 (i-1b), ultimate premolar P4 and M1 extracted from BMNH2942A. The M2 was extracted from BMNH2942B. f, Left lower teeth (stereo pair photos) in occlusal view. Note the roots of deciduous incisors are entirely medial to roots of premolar or molars. The M2 was extracted from BMNH2942B, and all other teeth are from BMNH2942A. g, Right lower M2 from BMNH2942B. h, Right lower M2 (stereo images, occlusal view). i, Right lower teeth (stereo images) in occlusal view from BMNH2942A. j, Right lower teeth in medial view. Bones removed digitally to expose tooth roots and the unerupted replacing incisors, highlighting the prolonged replacement of incisors relative to the adult premolars and molars.

  5. Extended Data Figure 5: Dual mortar–pestle occlusion of Vilevolodon molars, in contrast to embrasure occlusion of Haramiyavia. (430 KB)

    a, Vilevolodon right mandible in oblique posterior view, highlighting the M2/M2 orientation and the attachment of the mandibular middle ear. b, Cusp pattern of upper molars (in translucent outline). Note that M2 is imbricated more lingually to M1. c, Cusp pattern of lower molars. Note that M2 is imbricated more lingually to M1, mirroring the imbrication of upper M2 to M1. d, Superimposition of upper M1–M2 (translucent outline) over lower M1–M2 (grey) for dual mortar–pestle occlusion of the upper and lower molars (see also Fig. 3). e, Oblique posterior view of right M2 and M2 in dual mortar–pestle occlusion: A1 cusp (A1 pestle) entering the M2 basin and the A1 cusp (A1 pestle) entering the M2 basin. f, Dual mortar–pestle occlusion of left M2 and M2 (solid surface models) in occlusal views. g, Dual mortar–pestle occlusion of right M2 and M2 surface models in medial (lingual) view (g1), posterior (distal) view (g2), lateral (labial) view (g3), and anterior (mesial) view (g4). h, Occlusion of right lower M1–M2 to right upper M1–M2 in lingual view, based on the best fit of opposing teeth. Note that the B4–B3 cusps of upper molars are always lingual to the lower lingual cusp row, but the B1 cusp is always labial to lower lingual cusp row and in the lower median furrow. i, Occlusion of right M1–M2 to right M1–M2 in labial view. Note that cusp A4 of the upper molar is always labial (lateral) to cusp A1 of the lower molar. The upper labial row A1–A4 occludes obliquely over lower labial row B1–B4. There is no longitudinal alignment of cusp rows between the opposite upper and lower teeth, as in Thomasia, Haramiyavia and Maiopatagium (Fig. 3). j, Haramiyidan Haramiyavia occlusal trajectory (blue, orthal occlusion; green, palinal movement). k, Haramiyavia: transition of orthal phase (blue) to palinal phase (green) occlusion. Cusp row A1–A4 of the lower molar is lingual to cusp row B1–B5 of the upper molars. l, Haramiyavia: full orthal occlusion of M3 with A1 cusp in embrasure between upper M2–M3, and cusp row A1–A4 lingual to cusps B1–B5.

  6. Extended Data Figure 6: Autapomorphic occlusal features of Vilevolodon, including two inferred cycles of the molar occlusal movement. (230 KB)

    The upper tooth row has four teeth and is longer than the lower tooth row of three teeth. The upper P3–P4 flexure forms a prominent angle of the occlusal planes between the two premolars. In cycle 1 when lower M1–M2 is able to make full contact with M1–M2, P4 and P3 have no contact. In cycle 2, P4/P3 can make full contact, and their occlusion occurs only during cycle 2. But during this cycle, when the primary cusp of the P4 makes full occlusal contact with P3, the lower M1 and M2 can only barely contact the upper M1 and M2 and cannot make full contact. Occlusal movement was simulated by animation of STL models (see Supplementary Video 1) and inferred by maximal fit of upper and lower tooth occlusal surfaces. Blue arrows indicate the path of cusp A1 of the lower premolar. The blue band in the central column shows the differences in the position of maximum intercuspation (centric occlusion of the P4 primary cusp) between cycle 1 (above) and cycle 2 (below).

  7. Extended Data Figure 7: Dental and mandibular morphologies of eleutherodontids and extant frugivorous bats Sturnira and Artibeus (Chiroptera, Phyllostomidae), Hypsignathus (Megachiroptera: Pteropodidae), and frugivorous/omnivorous primate Daubentonia. (456 KB)

    ac, Vilevolodon left M2 (a) and left P4 (b, c). df, Lower teeth of frugivorous bats: Sturnira lilium lilium (FMNH105870) (d); Artibeus jamaicensis (FMNH30776) (e); and Hypsignathus monstrosus (University of Chicago Biological Sciences Division Teaching Collection) (f). g, Megaconus right mandible in lateral view, highlighting the zigzag tooth row profile9. h, Vilevolodon right mandible in lateral view, highlighting the zigzag tooth row profile. i, Hypsignathus mandible in lateral view. j, Artibeus mandible in lateral view, highlighting the zigzag tooth row profile. k, Primate Daubentonia (FMNH15529) right mandible in lateral view. a, df, Similarities of lower molars of Vilevolodon with the basined lower molar talonids of frugivorous bats Artibeus, Sturnira and Hypsignathus. Eleutherodontids are particularly similar to the highly creased talonid basins of Artibeus, whereas Maiopatagium is more similar to Sturnira and pteropodid bats in molar pattern. b–e, f, Similarities of lower premolar of Vilevolodon to those of Sturnira and Hypsignathus. g, h, Similarity of the zigzag tooth row profile of lower premolars and molars of Megaconus and eleutherodontids to those of frugivorous bats. g, h, k, Similarity in mandibles of eleutherodonts and the primate Daubentonia (FMNH15529), which is frugivorous and insectivorous.

  8. Extended Data Figure 8: Vilevolodon middle ear structure preserved in holotype (BMNH2942) and interpretive reconstruction. (530 KB)

    a, BMNH2942A in approximately ventral view. External and exposed aspect of left petrosal (partial) and middle ear bones as preserved. The dentary and cervical vertebrae are digitally removed so that the view of the middle ear features is unobstructed. b, BMNH2942A ‘internal’ view (underside of the fossil in matrix), showing the petrosal (partially segmented) and middle ear bones. c, Interpretative reconstruction of the middle ear bones of Vilevolodon, based on middle ear bones exposed in CT scans and 3D segmentation. d, A composite reconstruction of the middle ear of eutriconodonts. e, Cynodont Kayentatherium (redrawn from ref. 30). f, Pre-mammaliaform cynodont Cynognathus middle ear (redrawn from ref. 31) Note 1 to readers: the gracile ectotympanic (reflected lamina of angular), and this gracile and relatively straight morphology is similar to the reflected lamina of several non-mammalian cynodonts. The angle of attachment to the anterior–posterior limbs of ectotympanic (homologue of the angular) is slightly distorted by fossilization. Notes 2 and 3 to readers: the manubrial portion of the malleus (Note 2) is partly attached to the malleus body (Note 3), but there is a gap between these two structures owing to imperfect preservation. Note 4: the posterior end of Meckel’s cartilage is attached to a large sliver of bone. The latter is tentatively interpreted to be a surangular, which in non-mammalian cynodonts is typically parallel to the gonial part (the prearticular) of the malleus (although separated from the latter). There appears to be a suture between the prearticular part of the malleus, and the putative surangular. We infer that the Meckel’s cartilage and the ectotympanic, as preserved in close association, contact each other loosely. But they became slightly separated from each other during fossilization. We reconstructed these two elements as contiguous, as in cynodonts and other mammaliaforms, as demonstrated in c.

  9. Extended Data Figure 9: Comparative morphology of the mandibles and partial mammalian middle ears of extinct Mesozoic mammaliaforms. (438 KB)

    a, Morganucodon and several mammaliaforms possess plesiomorphic mandibular middle ear of cynodonts (MMEC). The ectotympanic (homologue of the angular) is nestled in the angular concavity of the mandible. The surangular and the gonial part (homologue of the prearticular) of the malleus make full contact with a broad postdentary trough that extends posteriorly to near the dentary condyle under the medial ridge. The Meckel’s element (ossified cartilage) starts from the postdentary trough, passes anteriorly below the mandibular foramen, and extends further anteriorly into a long Meckel’s sulcus below the mandibular tooth row. b, Eleutherodontid Vilevolodon shows a mandibular connection of the middle ear, although much more reduced (antero-posteriorly shortened) than those of other mammaliaforms. The anterior limb of the ectotympanic and the gonial part (prearticular) of the malleus are nestled in a reduced postdentary trough on the internal aspect of the inflected mandibular angle. The groove occupies the identical position of the postdentary trough in the angular region of other mammaliaforms. However, it is narrower and much shorter than the postdentary trough of other mammaliaforms. The trough is shorter (an apomorphic trait) and does not extend posteriorly to the dentary condyle, in contrast to the postdentary trough in other mammaliaforms. The short and tapering Meckel’s cartilage ends below the mandibular foramen, but it does not extend further anteriorly to the mandibular body below the tooth row. The Meckel’s element and its corresponding Meckel’s sulcus are much shorter than those of other mammaliaforms, eutriconodonts and spalacotherioids. The gonial portion of the malleus (and the putative surangular) also contacts the ventral margin of the mandible. We infer that, in Vilevolodon, the ectotympanic (retroarticular process) and the manubrial part of the malleus are rotated medially and horizontally, away from the mandible (b3), as in eutriconodonts. c, Eutriconodont Yanoconodon, highlighting the massive ossified Meckel’s cartilage that connects the middle ear to the mandible. Owing to the curvature and mid-length bending of Meckel’s element, the ectotympanic and malleal manubrium rotated medially and away from the mandible32.

Supplementary information

Video

  1. Video 1: Video 1: Analysis of tooth occlusion of Vilevolodon by animation of STL tooth models from CT scan (21.23 MB, Download)
    Part 1: Comparison of distinctive cycle 1 versus cycle 2 of premolars and molars: cycle 1 has full contacts of molars and incisors, but no contact between P3/p4; cycle 2 has full contact of P3/p4 but minimal or no contacts of molars and incisors. Part 2: Molar cusp and basin identification, and cycle 1 of occlusal movement of M1/m1 and M2/m2 during (see Fig. S6). During occlusal cycle 2, the P3/p4 are the only teeth that can fully occlude; the dual mortar-pestle contacts of upper and lower molars do not occur. Thus, cycle 2 is not shown here in animation analysis Part 2 of the video. For better clarity, the un-erupted deciduous incisors are not shown in the video.

PDF files

  1. Supplementary Information (4.9 MB)

    This file contains Supplementary Information parts A-Q.

  2. Reporting Summary (67 KB)

Additional data