LETTER doi:10.1038/nature12424 Evolutionary origins of the avian brain Amy M. Balanoff1,2{, Gabe S. Bever1,3, Timothy B. Rowe4 & Mark A. Norell1 Features that were once considered exclusive to modern birds, such intermediate between these early forms and modern birds3,4. Our new as feathers and a furcula, are now known to have first appeared in data indicate that the relative size of the cranial cavity of Archaeopteryx non-avian dinosaurs1. However, relatively little is known of the early is reflective of a more generalized maniraptoran volumetric signa- evolutionary history of the hyperinflated brain that distinguishes ture and in several instances is actually smaller than that of other birds from other living reptiles and provides the important neuro- non-avian dinosaurs. Thus, bird-like encephalization indicesevolved logical capablities required by flight2. Here we use high-resolution multiple times, supporting the conclusion that if Archaeopteryx computed tomography to estimate and compare cranial volumes of had the neurological capabilities required of flight, so did at least extant birds, the early avialan Archaeopteryx lithographica, and a some other non-avian maniraptorans. This is congruent with recent number of non-avian maniraptoran dinosaurs that are phylogeneti- findings that avialans were not unique among maniraptorans in cally close to the origins of both Avialae and avian flight. Previous work their ability to fly in some form5,6. established that avian cerebral expansion began early in theropod Birds are distinct among living reptiles in the degree to which their history and that the cranial cavity of Archaeopteryx was volumetrically brains, particularly their forebrains, are expanded relative to body size. g a Acrocanthosaurus atokensis Tyrannosaurus rex Alioramus altai Struthiomimus Coelurosauria Shuvuuia deserti Incisivosaurus gauthieri Oviraptorosauria Conchoraptor gracilis b Citipati osmolskae Maniraptora Khaan mckennai Deinonychosauria Dromaeosauridae Troodontidae Archaeopteryx lithographica Paraves Ratitae c Avialae Galliformes Anseriformes Aves Gruiformes Falconiformes Neoaves Pelecaniformes Procellariformes d Podicipediformes Gaviiformes f Psittaciformes Columbiformes Caprimulgiformes Coraciiformes e Piciformes Figure 1 | Coelurosaur phylogeny and partitioned endocranial casts. bulbs (orange), cerebrum (green), optic lobes (pink), cerebellum (blue) and a–e, Endocasts of Citipati osmolskae (IGM 100/978) (a), unnamed troodontid brain stem (yellow). Endocasts are not scaled to size. f, Sagittally sectioned skull (IGM 100/1126) (b), Archaeopteryx lithographica (BMNH 37001) (c), Struthio of Phaethon rubricauda with osteological landmarks highlighted to correspond camelus (ostrich) (d), and Melanerpes aurifrons (woodpecker) (e) divided into to the regions shown in the endocasts. g, Phylogeny of included taxa. Proposed neuroanatomical partitions based on homologous osteological landmarks episodes of encephalization are indicated by changes in colour. Phylogeny using computed tomography data. Partitions roughly correlate to the olfactory adapted from ref. 30. 1American Museum of Natural History, Division of Paleontology, New York, New York 10024, USA. 2Columbia University, Department of Earth and Environmental Sciences, New York, New York 10027, USA. 3New York Institute of Technology, College of Osteopathic Medicine, Old Westbury, New York 11568, USA. 4Jackson School of Geosciences, The University of Texas at Austin, Austin, Texas 78712, USA. {Present address: Department of Anatomical Sciences, Stony Brook University School of Medicine, Stony Brook, New York 11794, USA. 5 SEPTEMBER 2013 | VOL 501 | NATURE | 93 ©2013 Macmillan Publishers Limited. All rights reserved RESEARCH LETTER This index of encephalization ranges from six to eleven times higher in a 7–9 birds than other groups , and comparably large indices are known Tyrannosaurus rex 10–12 only among mammals . The hyperinflated forebrains of birds and 2.4 Acrocanthosaurus atokensis 11,13 mammals evolved independently , possibly in response to different 2.0 sensory cues; derived olfactory capabilities versus enhanced visual Alioramus altai 1.6 acuity14,15. Details of this neuroanatomical elaboration were recently Zanabazar junior Citipati osmolskae explicated for the mammalian side of the tree based largely on fossil 1.2 Conchoraptor Khaan mckennai evidence from the latter portion of the stem and the early history of the 0.8 gracilis 16 crown . No correspondingly comprehensive study exists for birds, 0.4 IGM 100/1126 Tsaagan mangas despite broad interest in the relationship between brain size and struc- Archaeopteryx lithographica 0.0 2 (log) Endocranial volume ture, cognitive ability, and the origin of avian flight . Shuvuuia deserti The volumetric expansion of the avian endocranium began relatively –0.4 early in theropod evolution4,17–19, and the early avialan Archaeopteryx –0.8 lithographica is volumetrically intermediate between those of more –1.8 –1.2 –0.6 0.0 0.6 1.2 1.8 2.4 3.0 3.6 3,4 basal theropods (for example, tyrannosaurs) and crown birds .What Body size (log) remains unclear is whether the Archaeopteryx endocranium will con- tinue to occupy a uniquely intermediate space between non-avialan theropods and crown birds once additional endocranial features, some b with neurological implications for flight, are sampled from a wider 2.4 range of bird-like, non-avian theropods. This is particularly relevant Tyrannosaurus rex considering that recent studies argued, first, that avialans are not uni- 2.0 Alvarezsauridae Oviraptorosauria Troodontidae Dromaeosauridae Archaeopteryx birds Crown que among maniraptorans in their ability to fly in some form20, and 1.6 second, that Archaeopteryx is more closely related to dromaeosaurs 1.2 Zanabazar junior Alioramus altai and troodontids than to modern birds21. 0.8 Citipati osmolskae Khaan mckennai We tested the relative position of Archaeopteryx in the evolution of Conchoraptor avian endocranial space using comparative volumetric analyses. Volu- 0.4 gracilis IGM 100/1126 Tsaagan mangas Cerebrum volume (log) volume Cerebrum 0.0 mes were obtained from digital endocasts constructed from computed Archaeopteryx lithographica tomography data sets for a diversity of crown and stem avians (Sup- –0.4 plementary Table 1). Our study differs from previous efforts in that we Shuvuuia deserti sampled those theropod lineages most closely related to Avialae; Troo- –0.8 dontidae, Dromaeosauridae, Oviraptorosauria and Alvarezsauridae –1.8 –1.2 –0.6 0.0 0.6 1.2 1.8 2.4 3.0 3.6 (Fig. 1). In addition to considering the relationship between total endo- Body size (log) cranial volume and body size, we also divided the endocasts into volu- Figure 2 | Bivariate plots of log-transformed body-mass data. a, b,Body metric partitions that estimate the major neuroanatomical regions, mass (kg) plotted against total endocranial volume (cm3)(a) and cerebral including the olfactory bulbs, cerebrum, optic lobes, cerebellum and volume (b). Crown birds display apomorphically high endocranial and cerebral brain stem (Fig. 1; Supplementary Table 2). This partitioning, accomp- volumes with respect to body size. Colours indicate crown birds (blue), non- lished using homologous osteological landmarks, enabled us to examine maniraptoran theropods (white), Shuvuuia deserti (purple), oviraptorosaurs how the volumetric signature of different endocranial regions evolved (red), deinonychosaurs (yellow), Archaeopteryx lithographica (green). Reduced in relation to total body size, total endocranial volume, and to one ano- major-axis regression line for entire sample (solid line), crown birds (large dashes), and non-avian theropods (small dashes). Regression statistics given in ther. This approach allows detection of previously unrecognizable evo- Supplementary Table 3. lutionary complexity. Our analysis of total endocranial volume relative to body size (esti- 4,9 mated based on femur length22), recovered the apomorphically high avian stem . The relatively high correlations between total volume volumetric signature for the avian crown (Fig. 2; Supplementary and partitioned volumes (Fig. 3 and Supplementary Fig. 2) indicate Tables 2 and 3)3,4,10,23, but failed to recover Archaeopteryx in a uniquely that each region, with the exception of the olfactory bulbs, expanded transitional position between non-avialan maniraptorans and crown along the avian stem. This suggests no notable structural constraint on birds. Several oviraptorosaurs and the troodontids Zanabazar junior total endocranial volume. If such a constraint were in place, then selection- and IGM 100/1126, all have relative endocranial volumes that fall driven cerebral expansion would be expected to occur at the volume- between the values of Archaeopteryx and crown birds (Fig. 2). The tric expense of one or more of the other regions. The fact that these same basic pattern was recovered when cerebral volume was compared regions are expanding together attests to the high responsive potential 24 to body size. Thus, the total endocranial and cerebral volumes of Arch- of the surrounding skeleton to the tissues they envelop . aeopteryx relative to body size are not uniquely avian but reflect ple- Principal components analysis of the five partitions divided by total siomorphic values expected of a non-avian maniraptoran. Even the endocranial volume reveals complete volumetric separation between uniquely derived signature of crown birds is lost when volumes of Paraves and Oviraptorosauria (Fig. 4). Principal