Introduction

The evolutionary origin of highly encephalized brains in modern birds is not only of interest to avian neuroscience but also has implications for how large brains evolve across vertebrates, including humans. Disentangling how the brain of crown birds (“Aves”) evolved requires study of brain anatomy ancestral to crown birds—that of non-avian dinosaurs. Using skulls and cephalic endocasts (internal molds of the braincase) as proxies for brain shape and size, previous studies have revealed that the evolutionary assembly of the avian brain form proceeded in a mosaic and homoplastic manner from the ancestral dinosaurian pattern1,2,3,4,5,6,7,8,9,10,11. While the tempo and mode of avian brain evolution is well established by recent studies, the mechanism underlying major neuroanatomical transformation along the dinosaur-bird transition remains uncertain3,12. Previous studies have proposed that a mixture of heterochronic processes within the cranium, particularly progenesis (truncation of the ancestral developmental period) and accelerated peramorphosis (evolutionarily driven increases in developmental rates), may underlie major changes to the brain architecture from ancestral archosaurian forms to crown birds3,11,12,13,14,15,16. However, a good developmental series of neuroanatomical morphologies for non-avian dinosaurs has been lacking to date; and crucial evidence is needed for a reliable inference of evolutionary shifts in brain development along the dinosaur-bird transition.

In this study, we examine an ontogenetic series of endocasts of the ornithischian dinosaur Psittacosaurus lujiatunesis, spanning perinatal to somatically mature individuals (<1-, 7-, 8-, and 10-year-old individuals) that were recently described17,18 (Fig. 1a; Supplementary Fig. 1). This allows for an unprecedented look into how the neuroanatomy of a non-avialan dinosaur species changes with age. We incorporate this developmental dataset into an interspecific dataset of archosaurian endocranial morphology which also includes postnatal ontogenetic series of the domestic chicken (Gallus gallus domesticus) and American alligator (Alligator mississippiensis) as an outgroup3 (Fig. 1a, b). Moreover, we include immature specimens of additional non-avialan dinosaurs17,18,19,20,21 (Lambeosaurus, Corythosaurus, tyrannosaurids, Zanabazar, and an unnamed troodontid; Supplementary Data 1) to further define the overall developmental axis of brain morphology in non-avialan dinosaurs (i.e., excluding Archaeopteryx considered here to be a basal avialan). Using a landmark-based geometric morphometric characterization of endocranial shape (Fig. 1c; Supplementary Table 1), the study aims to investigate how the magnitude and direction of developmental shape changes in brain morphology in non-avian dinosaurs compare to (i) the domestic chicken and the American alligator and (ii) the principal axis of brain shape evolution along the dinosaur-bird transition to evaluate whether heterochronic processes, particularly progenesis, underlie the origins of the highly derived avian brain. In our reporting, endocranial regions are referred to by their underlying soft-tissue features (i.e., “cerebrum” rather than “impression of the cerebrum”).

Fig. 1: Endocranial and landmark sampling for this study.
figure 1

a Endocranial reconstructions of neonate, intermediate, and mature specimens of Gallus, Psittacosaurus, and Alligator in lateral view. The intermediate ontogenetic representatives illustrate morphological transitions, not equivalent growth stages across taxa. Scale bars equal 5 mm. b time-calibrated phylogeny showing the taxonomic sampling of this study, where branches are color coded by major groups. Asterisks indicate taxa represented by immature specimens. c landmark scheme projected on the endocranial reconstruction of Psittacosaurus hatchling (IVPP V15451), where red, yellow, and blue coordinate points denote fixed, curve, and surface semi-landmarks in oblique (top), lateral (middle), and ventral (bottom) views. The numbers correspond to those assigned to discrete landmarks (Supplementary Table 1).

Full size image

Here, we show that the endocranial form of non-avian dinosaurs developed and evolved uniquely when compared to crown birds and Alligator mississippiensis. We demonstrate that ornithischians and non-avialan theropod dinosaurs share a common shape-to-size relationship that guides their brain development and evolution. Regional morphometric analyses of endocranial developmental trajectories reveal that non-avian dinosaurs collectively exhibit a mosaic of avian and crocodilian modes of allometric trajectories wherein the cerebrum and optic lobe regions of immature non-avialan dinosaurs resemble those of extant birds whereas the brainstem region follows the trajectory of the crocodilians. Our results thus support a paedomorphic trend occurring along the dinosaur-bird transition that is localized to some but not all regions of the brain.

Results

Endocranial shape morphospaces

To visualize total and regional endocranial shape variation, morphospaces were constructed based on principal components (PC) axes of endocast shape data (Fig. 2). In the morphospace of total endocranial shape, lower PC1 scores are associated with posterolaterally expanded cerebra, ventrally oriented optic lobes, anteroposteriorly compact hindbrains, and higher dorsoventral flexion with respect to both cephalic and pontine flexures as defined by Lautenschlager and Hübner22. Lower PC2 scores correspond to elongated cerebra, dorsoventrally extended optic lobes, anteroposteriorly shortened cerebellum, and greater dorsoventral flexion along the brainstem. As expected, total endocast shape data closely reflect the results found in previous geometric morphometric work where crown birds, alligators, and non-avian dinosaurs individually form clusters along PC1, with Archaeopteryx positioned between crown birds and non-avialan archosaurs along PC1 (Fig. 2a)3. Ornithischian and non-avialan theropod dinosaurs overlap with each other, and mature individuals of both groups strongly overlap with mature endocranial shape of Alligator along PC1 and PC2. With the exception of 7-year-old Psittacosaurus (IVPP V12731), endocasts of immature non-avialan dinosaurs generally exhibit lower PC1 scores closer to the crown bird cluster. The perinatal Psittacosaurus (IVPP V15451) and juvenile troodontid (IGM 100/1126), as well as the small-bodied, purportedly immature tyrannosaurid specimen (CMNH 7541), occupy an intermediary position between crown birds and non-avialan archosaurs along PC1 (Fig. 2a). All other juvenile taxa maintain similar PC2 scores as CMNH 7541 but have higher PC1 scores. Notably, the perinatal Psittacosaurus (IVPP V15451) and juvenile troodontid (IGM 100/1126), overlap in PC2 values as neo- and perinatal Gallus (Fig. 2a).

Fig. 2: Endocranial shape variation in extant birds, non-avian dinosaurs, and American alligator.
figure 2

Morphospaces constructed from first two principal components (PC) axes of (a) total endocast shape (inset images along axes illustrate extremes of shape changes along PC1 and PC2 axes; green, blue, and purple arrows show the general ontogenetic direction of Alligator, Psittacosaurus, and Gallus, respectively); b cerebrum shape; c cerebellum shape; d optic lobe shape; e brainstem shape. The regional shape data were extracted by performing local Procrustes alignment within each region.

Full size image

During development, the immature non-avialan dinosaur endocasts exhibit a posterolaterally expanded cerebrum, more ventrally oriented optic lobes, anteroposteriorly compressed hindbrain, and greater degree of dorsoventral flexion. In the case of IVPP V15451 and IGM 100/1126 (perinate Psittacosaurus and a juvenile troodontid, respectively), the anatomical arrangement of the endocast more closely resembles crown birds than the arrangement found in a somatically immature specimen of the early avialan Archaeopteryx does along PC1. With age, the morphological layout of the non-avialan dinosaur endocast trends towards that of Alligator, where the mature individuals of non-avialan dinosaurs and Alligator converge in their ontogenetic trajectories. This is to say that during postnatal development, the cerebrum narrows anteriorly relative to the rest of the endocast, the optic lobes become less bulbous, the hindbrain expands posterolaterally, and the degree of dorsoventral flexion decreases. P. lujiatunensis follows this trend between the perinatal and mature stages (8- and 10-year old) with the 7-year-old showing endocasts that resemble endocranial shape of subadult Alligator (Fig. 2a). The seemingly strange position of 7-year-old Psittacosaurus may be due to the dorsoventrally shallower morphology compared to the three other Psittacosaurus endocasts, which may be due to moderate degree of deformation (Supplementary Fig. 1). Juvenile hadrosaurs, such as the juvenile Lambeosaurus (ROM 758) and Corythosaurus (ROM 759), generally plot more closely with alligators than crown birds, but it is unclear, with current sampling, if perinatal individuals would plot more closely to crown birds as is the case with Psittacosaurus.

In general, the morphospace of cerebrum shape (Fig. 2b) generally mirrors the pattern observed in total endocranial shape, where the alligator, non-avian dinosaurs, and crown birds form clusters with partial overlap along the PC1 axis. This axis encompasses the spectrum from the more globular cerebrum in modern birds to more anteroposteriorly elongated profile in mature alligators. The orientation of the PC2 axis corresponds closely to the ontogenetic series of Alligator, which encompasses an anteroposteriorly constricted cerebrum in immature individuals to a more elongated profile in mature alligators. Overall, non-avialan dinosaurs group together in the area of morphospace between mature alligators and crown birds. Among non-avialan dinosaurs, the cerebrum shape of immature specimens, troodontids, and tyrannosaurids (such as the taxonomically problematic ?Nanotyrannus specimen CMNH 7541) group closest to crown birds (Fig. 2b). Endocranial shape of relatively mature ornithischian dinosaurs, including that of hadrosaurs and Psittacosaurus, overlap with that of mature alligators (Fig. 2b).

In the cerebellum shape morphospace, crown birds and Archaeopteryx form a group that is almost completely separated from alligators along PC2 (Fig. 2c). The PC1 axis is primarily associated with the anteroposterior extent and dorsoventral depth of the cerebellum. The PC2 axis mostly corresponds to the dorsoventral curvature. Non-avialan dinosaurs overlap both crown birds and alligators in cerebellar shape along both PC1 and PC2 axes. Relatively mature specimens of tyrannosaurids, troodontids, and hadrosaurs strongly overlap with relatively mature alligators. Interestingly, all the Psittacosaurus specimens lie among crown birds regardless of ontogenetic stage, although this similarity is likely due to the presence of a prominent dorsal eminence, thought to represent space occupied by non-neuronal soft tissues (e.g., venous sinus, meninges), on the cerebellum (Fig. 1a; Supplementary Fig. 1). Apart from the juvenile hadrosaurs, Lambeosaurus (ROM 758) and Corythosaurus (ROM 759), all juvenile specimens of troodontids, the small-bodied tyrannosaurid specimen CMNH 7541, and other ornithischian dinosaurs, overlap with crown birds in PC2 values.

In optic lobe shape morphospace (Fig. 2d), the distinctions between alligators, non-avialan dinosaurs, and crown birds are more apparent, particularly along PC1, which documents the range from the anteroposteriorly constricted and dorsoventrally elongated optic lobes in alligators to the hemispherical optic lobes in modern birds and some non-avialan dinosaurs. PC2 encompasses the range from convex to topologically flat optic lobes. Alligators and crown birds overlap along PC2 due to their convex optic lobes. Optic lobes of some non-avialan dinosaurs closely resemble those of alligators and crown birds. For example, the subadult Psittacosaurus (PKUP V1054) lies very close to alligators (which could be due to taphonomic deformation as noted above) and away from other congeneric specimens in PC1 values. The perinatal Psittacosaurus (IVPP V15451), the oviraptorid Khaan (IGM 100/973), and Archaeopteryx are adjacent to the area of morphospace occupied by crown birds.

The morphospace for brainstem shape visually separates crown birds and all other taxa sampled in this study along PC2 (Fig. 2e). The PC1 axis largely documents the ontogenetic shape changes in alligators and chickens, from a mediolaterally expanded and dorsoventrally curved brainstem in perinatal specimens to anteroposteriorly elongated and weaker dorsoventral flexion in adults. Along PC2, crown birds and the chicken ontogenetic series cluster together, while non-avialan dinosaurs overlap considerably with adult and subadult alligators. More immature alligators were grouped with Archaeopteryx, relatively immature Psittacosaurus specimens (IVPP V15451, PKUP V1053, and PKUP V1054), and the immature troodontid IGM 100/1126 (Fig. 2e).

Allometric trajectories

To compare shape-to-size relationships, we visualized and tested how non-avialan dinosaur endocrania differed from allometric trends in crown birds, together with developmental trajectories of Gallus and Alligator. For visualization of allometric trends, PC1 of residuals from the common allometric trend (RSC1) were plotted against scores along the allometric trendline (CAC: common allometric component) for all endocrania sampled (Fig. 3). As in previous work5, total shape variation among the endocrania of crown birds is allometrically divergent from non-avialan dinosaurs and Alligator (Fig. 3a). As expected, crown birds occupy a restricted size range along the CAC axis but exhibit a high degree of shape variation along the RSC1 axis. The ontogenetic series of chickens overlaps the interspecific trajectories of the other crown birds used in this study (Fig. 3a). Whereas immature non-avialan dinosaurs exhibit endocranial shapes that are distinct from both crown birds and alligators, more mature specimens of ornithischian and non-avialan theropod dinosaurs converge towards the endocranial shape of mature Alligator. More immature Alligator specimens deviate further from crown birds along RSC1.

Fig. 3: Allometric trajectories constructed from common allometric component (CAC) and PC1 of residual shape from CAC.
figure 3

a total endocast; b cerebrum shape; c cerebellum shape; d optic lobe shape; e brainstem shape. The lines and grey bands represent regression lines for respective taxonomic group and their 95% confidence intervals, respectively. The regional shape data were extracted by performing local Procrustes alignment within each region. When appropriate, green, blue, and purple arrows show the general ontogenetic direction of Alligator, Psittacosaurus, and Gallus, respectively. Source data and code are provided as a Source Data file.

Full size image

The allometric trajectory of cerebrum shape in crown birds, along with the ontogenetic series of chickens, are separate from the entire ontogenetic series of Alligator (Fig. 3b). With the exception of perinatal Psittacosaurus (IVPP V15451), sampled ornithischian taxa exhibit cerebrum shapes that overlie the relatively mature alligators. Notably, the perinatal Psittacosaurus specimen (IVPP V15451) aligns with crown birds in RSC1 values, whereas the juvenile Lambeosaurus (ROM 758), juvenile Corythosaurus (ROM 159) and more mature ornithischian dinosaurs have RSC1 values that are closer to adult Alligator. Non-avialan theropods span the range of endocranial shape of alligators and modern birds, where the oviraptorid Citipati and the tyrannosaurid CMNH 7541 are within the area occupied by some of the crown birds, such as Phalacrocorax (cormorant) and Raphus (dodo). Hadrosaurs have a larger RSC1 score than the other non-avialan dinosaur taxa sampled, with subadult Psittacosaurus having the lowest. Alligators span the most along RSC1, with hatchlings having large RSC1 scores that decrease notably as they mature to adulthood (Fig. 3b).

Similar to the cerebrum, allometric trajectories of the alligator and crown bird cerebella are distinct, with non-avialan dinosaurs overlapping with these two taxa (Fig. 3c). Psittacosaurus, immature theropods, and non-avialan maniraptoran dinosaurs group within or near crown birds, with the juvenile Lambeosaurus (ROM 758) and more mature non-avialan dinosaurs more closely resembling adult Alligator (Fig. 3c). Adult and subadult Psittacosaurus plot with crown birds along with Tyrannosaurus (AMNH 5029) and Gorgosaurus (ROM 1247) specimens, while the juvenile and subadult hadrosaurs align more closely with Alligator (Fig. 3c). It is worth noting that non-avialan dinosaur taxa positioned close to crown birds, including Zanabazar and CMNH 7541, show prominent dorsal eminence in the endocast.

For the optic lobe, crown birds and Archaeopteryx group together along the RSC1 axis (Fig. 3d). Non-avialan dinosaurs loosely group together, with the oviraptorosaurs Incisivosaurus and Khaan closely resembling the optic lobe shape of crown birds and Archaeopteryx and the subadult Psittacosaurus (PKUP V1054) positioned with adult Alligator (Fig. 3d). The perinatal Psittacosaurus (IVPP V15451) and immature troodontids Zanabazar and unnamed troodontid (IGM 100/1126) are close to the RSC1 values of crown birds than other endocasts from non-avialan dinosaurs. Based on the plot, it is uncertain whether non-avialan dinosaurs align more closely with the allometric trajectories of alligators and/or crown birds.

Allometric trajectories of brainstem shape of crown birds and Gallus ontogeny are mostly distinct from ontogenetic series of Alligator but converge throughout postnatal ontogeny of the brainstem (Fig. 3e). Archaeopteryx and non-avialan dinosaurs of all ages align with the ontogenetic series of Alligator. In particular, the perinatal Psittacosaurus (IVPP V15451) and juvenile troodontid IGM 100/1126 follow the trendline of Alligator ontogeny, away from the allometric trajectory governing scaling relationships across crown birds and chicken development. As such, the results suggest that both ornithischian and non-avialan theropod dinosaurs resembled an alligator-grade scaling relationship in brainstem shape.

In addition to visual comparison of allometric trajectories, we also statistically tested the differences in allometric trajectories. The non-avialan dinosaur with the most complete ontogenetic series of cranial endocasts in our study, P. lujiatunensis, was compared against allometric trends in chickens and alligators and was found to have a distinct allometric trajectory. This result is confirmed by MANOVA on full Procrustes coordinate data against log-transformed centroid size and trajectory analyses for mean angle differences (Supplementary Table 2), when compared to Gallus (MANOVA: R2 = 0.352; P < 0.0001; trajectory analysis: mean distance = 0.218, P < 0.0001; angle difference = 16.8°, P < 0.0001) and Alligator (MANOVA: R2 = 0.210; P < 0.0001; trajectory analysis: mean distance = 0.152, P < 0.0001; angle difference = 11.6°, P < 0.0001).

Discussion

Non-avialan dinosaurs exhibit distinct allometric trajectories

It has generally been assumed that most non-avian dinosaurs would show brain developmental patterns more akin to birds than crocodilians although poor ontogenetic sampling has made this difficult to test3. If true, this would further support the notion that the crocodilian-like ontogenetic sequence is the plesiomorphic mode among early archosaurs23,24,25. With ontogenetic series from a non-avialan dinosaur, our results indicate that dinosaurian brain ontogeny differs from both crown birds and crocodilians, showing a mixture of plesiomorphic and derived neuroanatomical trends. We further demonstrate that ornithischians, such as Psittacosaurus, and non-avialan theropods share statistically indistinguishable allometric trajectories. In both the PC morphospace and allometric trajectories, the results show that the ontogenetic trajectories of total endocast shape in non-avialan dinosaurs lie between those of crown birds and alligators (Figs. 2a, 3a). Indeed, they begin partially avian-like among perinatal and juvenile individuals and broadly converge to more alligator-grade endocranial morphology. Because this allometric trajectory occurs in both coelurosaurian and ornithischian non-avian dinosaurs, this postnatal trend in brain shape development presumably originated in their common ancestor in the Triassic after diverging from crocodile-line pseudosuchian archosaurs. This then identifies three major postnatal ontogenetic trajectories for endocast shape in archosaurs: the crocodilian style in which the brain becomes proportionately elongated anteroposteriorly post-hatching, the avian style which remains anteroposteriorly short and dorsoventrally flexed throughout postnatal development, and the non-avialan dinosaurian style which exhibits a mosaic of avian and crocodilian neuroanatomical features, as has been qualitatively noted interspecifically within clades5.

When individual endocast regions are analyzed, the degree of overlap in endocranial shape and allometric trajectories differ across regions. In cerebrum, cerebellum, and optic lobe shapes, non-avialan dinosaurs resemble more those of crown birds than alligators (Fig. 2b–d). However, this trend is reversed along the brainstem where non-avialan dinosaurs and alligators group together and apart from crown birds (Fig. 2e). Based on allometric trajectories, the cerebrum shape of mature and immature non-avialan dinosaurs align more with mature alligators than immature Alligator (Fig. 3b). Both PC and RSC1 morphospaces show that cerebellum shape of immature non-avialan theropod dinosaurs, plus perinatal Psittacosaurus, align with the allometric trajectory of crown birds, with specimens of mature non-avialan dinosaurs converging to the shape of mature alligators (Fig. 3c). Whether this is a shape difference in actual cerebellum tissue or surrounding structures, such as venous sinuses, is difficult to infer because the correspondence between the brain and endocranial cavity is lower in the hindbrain. Nevertheless, we believe that this region of the endocast signifies important structural changes occurring in the posterior section of the brain cavity. Besides the optic lobe shape being closer to, yet distinct from, crown birds along PC1, PC2, and RSC1 (Fig. 3d), the brainstem shape of ornithischian dinosaurs visually aligns with the allometric trajectory of alligators, whereas that of other non-avialan theropod dinosaurs, particularly immature specimens, aligns more closely with crown birds (Fig. 3e).

The regional analysis allows a more mosaic view of how non-avialan dinosaurs differ from both alligators and crown birds in the way endocranial morphology develops. The cerebrum, optic lobes, and possibly cerebellum regions of non-avialan dinosaurs display different degrees of resemblance to avian-like morphologies throughout development, whereas the brainstems of non-avialan dinosaurs more closely resemble those of alligators. The combination of heterogeneous developmental patterns between endocranial regions collectively creates a distinct endocast-wide developmental trajectory observed among non-avialan dinosaurs. Specifically, it seems that the crocodilian-style growth of the brainstem drives the unique morphological development of the non-avialan dinosaur endocranium, independent of locomotory differences between bipedalism and quadrupedalism.

Heterochronic trends in non-avialan dinosaur brain shape

Without meaningful developmental sampling of non-avialan dinosaurs, previous studies on endocranial shape could not reliably infer the evolutionary shifts in neuroanatomical development that occurred along the dinosaur-bird transition3. Previous work suggested that non-avialan dinosaurs had a unique postnatal neuroanatomical development in which the endocasts of juvenile specimens exhibited a more avian-like shape but then converged to a more crocodilian-like pattern with age3. The non-avialan dinosaurs used in this study tend to decrease their endocranial cephalic and pontine flexure angles with age, although not to the same degree of crocodilians. Ontogenetic variation in endocranial flexion has been noted by other authors12,26,27, but our study demonstrates that anatomical shifts also accompany the ontogenetic anteroposterior rearrangement of the endocast among some taxa as well. Crocodilians undergo this shift in endocranial flexion where their endocrania are more linear in lateral profile12,26,27. The endocranial shape of non-avialan dinosaurs exhibits transformation towards increased dorsoventral flexion early in their postnatal ontogeny but does not attain the near-linear layout of somatically mature crocodilians in maturity.

Our results suggest that paedomorphosis played a role in the total endocranial form and evolution of the avian brain where the form of the bird brain more closely resembles that of juvenile non-avialan dinosaurs. The shift from a non-avialan dinosaur endocranial form to that of an avian form is best explained through paedomorphosis rather than just miniaturization. Miniaturization of the body including the brain was a stepwise process that reduced the body mass along stem birds and with further reductions in body sizes occurring in early avialans28,29. And while the reduction of the brain size may be sustained and has been shown to occur at faster rates among bird-line theropods than other non-avialan dinosaur lineages30, miniaturization does not wholly explain the juvenilized non-avialan dinosaur neuroanatomy after the origin of Avialae (see also Torres et al. 11). This is demonstrated by the distinct allometric trajectories of crown birds and non-avialan dinosaurs, where differences in size alone cannot account for their brain shape evolution.

The paedomorphic features found in the avian brain are seen in all juvenile non-avialan dinosaurs and Archaeopteryx to varying degrees, further supporting paedomorphosis as at least a partial driver of avian endocranial evolution. Our work supports this hypothesis by demonstrating that all immature non-avialan theropods have a more posterolaterally expanded cerebrum, anteroposteriorly compressed hindbrains, and more sigmoidal lateral profile than their adult counterparts (i.e., all younger, immature specimens have lower PC1 scores than their mature counterparts). These features are also observed in the taxonomically contentious tyrannosaurid specimen, CMNH 7541, which has been considered a juvenile based on the bone surface texture and gross anatomy31 (but see32 for interpretation as a near-adult specimen). Regardless of its taxonomic or ontogenetic status, the endocast of CMNH 7541 shows traits associated with immaturity. This could be due to the specimen representing a juvenile individual or the taxon undergoing paedomorphism in both body size and endocranial morphology.

As in skull shape evolution from crocodilians to birds13,14, endocranial shape data suggest that non-avialan dinosaurs do not conform strictly to a single heterochronic process. Even with endocranial growth varying between non-avialan dinosaur taxa, all the specimens used in this study exhibit endocast shapes and allometric trajectories that are distinct from crown birds and crocodilians. Our finding of a uniquely intermediary endocranial ontogenetic trajectory in non-avian dinosaurs opens a question about whether the crocodilian development mode was the ancestral archosaur condition. If the Alligator condition is considered plesiomorphic, then the first major pulse of evolutionary changes towards avian-grade brain development may have occurred as early as the Triassic, before the phylogenetic divergence between Ornithischia and Theropoda. Alternatively, the dinosaurian condition may be ancestral to crown archosaurs, although support for this hypothesis would require extensive testing from a temporally and taxonomically broader dataset, including basal and pseudosuchian archosaurs. Furthermore, our results suggest that a secondary evolutionary pulse occurred wherein shared paedomorphic traits (i.e., posterolaterally expanded cerebra, bulbous optic lobes, and compressed hindbrains) amongst immature non-avian dinosaurs were further retained in avians.

Compared to previous studies, the inclusion of ontogenetic and interspecific sampling of non-avialan dinosaurs enables inferences to be made on the evolutionary scenario along the dinosaur-bird transition. The endocast shape data provide evidence that (i) ornithischian and non-avialan theropod dinosaurs exhibit a shared shape-to-size relationship that corresponds to their brain development and evolution. Regional analysis suggests that (ii) this unique allometric trajectory of non-avialan dinosaurs is due to acquisition of a derived developmental trajectory in the cerebrum, cerebellum, and optic lobe; and (iii) provides morphological evidence of the paedomorphic trend in the shape of the cerebrum and optic lobe along the lineage from early avialan to crown birds.

Methods

Specimens

We amassed 85 previously published endocranial reconstructions that span 52 species of non-avian and avian dinosaurs in addition to American alligators (Supplementary Data 1). The data include interspecific sampling of 36 crown birds, a juvenile Archaeopteryx33, and 19 non-avialan dinosaurs1,3,17,18,19,20,34, plus ontogenetic series of the domestic chicken (Gallus gallus; n = 14), the American alligator (Alligator mississippiensis; n = 14), and the ornithischian Psittacosaurus lujiatunensis3,17,35 (n = 4; Supplementary Fig. 1). Among non-avialan dinosaurs, several isolated immature specimens of both ornithischian and non-avian theropod dinosaurs, including Lambeosaurus (ROM 758) and troodontids Zanabazar and an unnamed specimen IGM 100/1126 were sampled for both ontogenetic and phylogenetic diversity19,20,21,36,37,38,39. The taxonomically ambiguous tyrannosaurid specimen, CMNH 7541, has been considered a relatively immature specimen within Tyrannosauridae40,41,42,43,44,45. The major results and conclusions of this study are not dependent on the taxonomic assignment of this specimen, but rather the ontogenetic stage of CMNH 7541. Considering that no specimen attributable to a tyrannosaurid similar to, and including, CMNH 7541 has preserved an external fundamental system that indicates the cessation of rapid growth and the onset of skeletal maturity32,40, we treat CMNH 7541 as an immature individual. The specimen and data acquisition protocols for Alligator and Gallus specimens were approved by Institutional Animal Care and Use Committee at respective institutions (Stony Brook IACUC Protocol #236370-1; Oklahoma State University Center for Health Sciences IACUC Protocol #2015-1; American Museum of Natural History 2014). Please refer to the references listed in Supplementary Data 1 for additional details about those specimens.

Imaging and endocranial reconstructions

Endocranial reconstructions were generated from computed tomography (CT) imaging of cranial specimens. Considering the incomplete nature of the PKUP 1053 and PKUP 1054 endocasts from previous study 18, they were reconstructed again for this study. We found that the original specimens and reconstructed models were of sufficient quality for morphometric analysis. These reconstructions were paired with two other Psittacosaurus specimens, IVPP V15451, and IVPP V12617, to reconstruct an ontogenetic series for the species P. lujiatunensis. The endocast of the tyrannosaurid Alioramus altai (IGM 100/1844) was retrodeformed using a local symmetrization algorithm46 implemented in the “Morpho” R package47 due to its sheared appearance. Because the combined data comprise multiple sources, the scanner, scanning parameters, and segmentation protocol naturally differ across datasets (Supplementary Data 1). Nevertheless, we considered variation in resulting shape data due to differences in these practices to be negligible compared to the large interspecific variation observed across archosaurs48. However, to adopt some standardization across endocranial reconstructions, we used the “QuickSmooth” function in Geomagic Wrap v2021 (3D Systems, Rock Hill, SC, USA) to remove cranial nerve and vascular projections from the main body of all the endocasts at the base.

Morphometric data

We adopted the same high-density 3D landmark scheme from previous studies to characterize the overall shape of endocasts and their major regions, including the cerebrum, optic lobe, cerebellum, and brainstem3,34,35 (Fig. 1c; Supplementary Table 1). The ‘patch’ tool in Landmark Editor v3.649 was used to place discrete landmarks that are anatomically defined, curve semi-landmarks that delimit regions, and surface semi-landmarks that capture the surface morphology within regions on both left and right sides of the endocast. We did not collect coordinate data from the olfactory tract and bulbs due to their incomplete preservation in the fossils, as well as their diminutive size in most crown birds. We acknowledge that endocasts are generally accurate representations of actual brain morphology among archosaurs35,50, but their correspondence is relatively low in the hindbrain due to greater presence of tissues surrounding the brain. This correspondence is lower in crocodilians27 and presumably in non-avialan dinosaurs, especially the presence of dorsal eminence in some endocasts19,20,51.

The coordinate data were subjected to generalized Procrustes alignment52,53 minimizing total bending energy, allowing semi-landmarks to slide on the mesh surface54,55 using the “slider3d” and “gpagen” functions in the “Morpho”56 and “geomorph”57 R packages. After alignment, (semi-)landmarks on the right side were removed to remove redundant shape information while minimizing artifacts caused by alignment of one-sided coordinate data of bilaterally symmetric structures58,59. We extracted centroid size from the pre-aligned coordinate data, in addition to the shape data to be analyzed comprising 14 discrete landmarks, 49 curve semi-landmarks, and 56 surface semi-landmarks (Supplementary Data 2, 3). For region-specific analyses, the size and shape of each region were extracted by calculating centroid size and performing localized alignment on a subset of landmarks that characterize each region.

Time-calibrated phylogeny

To perform phylogenetic comparative analysis, we constructed a time-calibrated phylogeny that includes all the sampled species. Because the crown group sampling is equivalent to a previous study3, fossil taxa were grafted to the existing tree file (additional details on the base tree construction3). These additional taxa included Corythosaurus, Hypacrosaurus, Psittacosaurus, Gorgosaurus, and Tyrannosaurus, which were incorporated into the existing phylogeny based on the mid-point of range of first occurrence age recorded on the Paleobiology Database (paleobiodb.org). When the maximum age of a species was identical to that of its clade, the age of the internal node was set to equally bisect the parent and descendent branches (“Equal Branching Method”)60.

Analysis

All statistical analyses for this project were performed in R v4.2.161. Patterns of endocast shape variation were visualized as morphospaces for the cerebrum, cerebellum, optic lobe, and brainstem regions of the endocasts and were constructed using scores from PCAs made from shape data. The functions physignal, procD.lm, and procD.pgls of the geomorph package were used to assess the level of phylogenetic signal, allometry, and evolutionary allometry, with 1000 pseudoreplications. The multivariate statistical tests outlined above were chosen due to their robusticity against type I errors and loss of power that often accompanies differences in specimen and landmark sampling62,63,64. Allometric trends were visualized by plotting the PC1 of residuals from the overall shape-to-size relationship against scores along this allometric relationship65. We used Morpho package’s “CAC” function to extract CAC scores and residuals from shape data. Statistical differences between endocranial shapes and allometric trajectories between clades were tested with the procD.lm function, where differences in clades were tested on data accounting for log-transformed centroid size. In addition, we used the “trajectory.analysis” function in the ‘RRPP” R package66 to test for differences between vectors of evolutionary and developmental shape change in total shape space.

Reporting summary

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