Author response:
Public Reviews:
Reviewer #1 (Public review):
Summary:
The study presents a novel analysis of MRI resources for 16 avian species, spanning major (though not all) clades and ecological niches. This is a significant step towards large-scale datasets on internal parcellation and long-range connectivity, central to evolutionary studies for understanding the evolution of the bird brain.
Strengths:
The integration of high-resolution T2-weighted and diffusion-weighted MRI with histological validation (Nissl and Luxol Fast Blue staining) provides a strong, cross-validated framework for studying avian brain anatomy. Data on long-range connectivity are particularly useful for understanding how relationships between brain components evolved. The approach is also scalable, allowing for more detailed evolutionary analyses compared to what is currently possible.
Weaknesses:
The sampling supports evidence of modular evolution in the bird brain, but it is limited for broad evolutionary claims, as the effects of sizes and phylogenies can be hard to disentangle without enough species per clade.
We appreciate the reviewer highlighting this limitation and agree that it warrants explicit consideration. Although the 16 species included in the current study encompass diverse avian lineages and ecological characteristics, our sampling was not designed to rigorously disentangle the effects of phylogeny and brain size within individual clades.
To improve the taxonomic coverage of our dataset, we plan to expand the MRI dataset in the revised manuscript by including additional specimens that are currently available to us. Specifically, we will acquire and analyze MRI data from the slaty-backed gull (Larus schistisagus), black-tailed gull (Larus crassirostris), budgerigar (Melopsittacus undulatus), and crow (Corvus sp.). In addition, we plan to incorporate publicly available MRI data from the ostrich (Struthio camelus). These additions will broaden the phylogenetic and anatomical coverage of the comparative dataset.
Nevertheless, we fully acknowledge the reviewer’s point that, even with these additional species, the number of species sampled within individual clades will remain insufficient to rigorously separate phylogenetic effects from size-related effects or to support broad generalizations across all avian lineages. We will therefore explicitly state this limitation in the revised manuscript and temper evolutionary claims that extend beyond what can be supported by the present taxonomic sampling.
Accordingly, we will more clearly position the primary contribution of this study as the establishment of an MRI-based comparative resource and analytical framework applicable across diverse avian species, rather than as a comprehensive phylogenetic test of avian brain evolution. Within this scope, we will present the observed variation among brain compartments as evidence consistent with mosaic/modular diversification, while carefully limiting the broader evolutionary interpretation of these patterns.
Tractography-based claims should be treated cautiously without sensitivity analyses. This is particularly important when comparing brains with different sizes and tissue properties.
The reviewer raises an important point regarding the interpretation of tractography-based comparisons. We agree that particular caution is required when comparing datasets across species that differ in brain size, tissue properties, and imaging characteristics.
In the revised manuscript, we will perform additional sensitivity analyses to evaluate the robustness of the major tractography-derived patterns. Specifically, we will examine the effects of varying the FA threshold used for tract reconstruction, rather than relying solely on the current threshold of 0.1, and assess whether the major reconstructed trajectory patterns and cross-species differences are robust to this parameter.
We will also re-analyze the available diffusion data using Generalized Q-Sampling Imaging (GQI) as an alternative reconstruction approach and compare the resulting trajectory patterns with those obtained using the current DTI-based analysis. We recognize that the ability to resolve complex fiber configurations depends on the underlying diffusion acquisition parameters, including b-value and spatial and angular resolution. We will therefore interpret these comparisons within the limitations of the currently available datasets, without implying that either reconstruction approach provides a definitive representation of the underlying fiber architecture.
In addition, we will provide a more detailed description of the relevant diffusion MRI acquisition parameters and spatial and angular resolution of the datasets so that potential technical differences among species can be more clearly evaluated. We will also discuss how such differences may affect cross-species tractography comparisons.
Finally, we will revise the interpretation of the tractography results throughout the manuscript to more clearly distinguish diffusion MRI-derived reconstructed trajectories from anatomically demonstrated neuronal connections. We will avoid treating reconstructed streamlines as direct evidence of anatomical connectivity and will explicitly discuss the potential for false-positive and false-negative tract reconstruction, as well as other limitations inherent to diffusion MRI tractography.
Existing literature is not acknowledged sufficiently. This makes some claims of novelty misleading, and prevents readers from understanding the current state of knowledge in this research area.
We fully agree with this assessment. Although substantial comparative neuroanatomical work has established important principles of avian brain evolution, the current manuscript does not sufficiently cite or incorporate this literature into the discussion. As a result, some statements regarding the novelty of the present study are overstated.
In the revised manuscript, we will substantially expand our discussion and citation of the relevant literature, including the extensive comparative studies of individual avian brain regions and brain subdivisions highlighted by the reviewers. We will revise the Introduction and Discussion accordingly to more accurately describe the current state of knowledge in comparative avian neuroanatomy and to clearly distinguish the contributions of previous studies from those of the present work.
We will also carefully revise statements regarding the novelty of our study. Rather than implying that our study provides the first comparative framework for examining internal avian brain organization, we will more appropriately emphasize its contribution as an MRI-based comparative resource that enables standardized visualization and analysis of internal brain anatomy and connectivity across diverse avian species.
Reviewer #2 (Public review):
This manuscript presents a comparative MRI dataset from 16 avian species and uses MRI and tractography to examine variation in brain organization across birds. The authors argue that their analyses support mosaic brain evolution and provide a framework for comparative neuroanatomy. Although the dataset represents a useful resource, particularly given the inclusion of understudied species like penguins, toucans, and hornbills, I have substantial concerns regarding the novelty of the study, the anatomical interpretation of the results, and the validity of the tractography analyses. In its current form, I do not believe the manuscript provides sufficient new biological insight to support many of its conclusions.
Major Concerns
(1) The authors repeatedly state that comparative neuroanatomical studies in birds have largely been unable to examine internal brain organization or "internal parcellation". This claim is inaccurate and reflects limited engagement with a substantial body of literature. For decades, comparative studies have examined variation in the size of major avian brain subdivisions as well as specific sensory, motor, and associative nuclei. For example, the extensive work of Andrew Iwaniuk and colleagues has documented variation in numerous brain regions across birds and related these differences to ecology, behavior, and sensory specialization (e.g., Gutierrez-Ibanez et al., 2009; Iwaniuk et al., 2006, 2008, 2010; Corfield et al., 2015). Other authors have also made important contributions in this area (e.g., Boire and Baron, 1994; Burish et al., 2004; Moore and DeVoogd, 2011, 2017). Importantly, previous work has already examined variation in major subdivisions of the avian brain using relatively standardized datasets (e.g., Iwaniuk et al., 2004; Iwaniuk and Hurd, 2005), including datasets that contain more species and greater taxonomic diversity than the current study. In other words, these studies have already provided detailed analyses of internal brain organization across broad taxonomic samples.
The manuscript should therefore be reframed as providing a new MRI-based resource rather than introducing the first comparative framework for studying internal avian brain organization. The current framing significantly overstates the novelty of the work.
We appreciate the reviewer drawing attention to this issue. Although a substantial body of comparative neuroanatomical work has already examined internal brain organization in birds, the current manuscript does not sufficiently cite or incorporate this literature into the discussion. As a consequence, some statements regarding the novelty of our study are overstated.
In the revised manuscript, we will expand our discussion and citation of previous work, including the studies highlighted by the reviewer on interspecific variation in major avian brain subdivisions, sensory and motor nuclei, and other anatomically defined brain regions. We will revise the Introduction and Discussion to more accurately represent the existing body of comparative avian neuroanatomy and to clarify how the present study complements and extends these established approaches.
Most importantly, we will reframe the manuscript so that its primary contribution is presented as the establishment of an MRI-based comparative resource for visualizing and quantitatively analyzing internal brain anatomy across diverse avian species. We will remove or revise statements implying that the present study provides the first comparative framework for examining internal avian brain organization. Instead, we will emphasize the complementary advantages of the MRI-based approach, particularly its ability to provide non-destructive three-dimensional visualization of internal brain structures in intact specimens and to enable comparison of multiple brain compartments within a common analytical framework across diverse avian species.
We believe that this revised framing will more accurately position the contribution of our study within the existing literature and clarify the specific value of the dataset.
(2) A second significant concern is the lack of anatomical specificity in the tractography analyses. The authors repeatedly refer to regions such as "anterior cortex," "dorsal cortex," and "temporal cortex." These terms are not standard anatomical designations in avian neuroanatomy and provide little information about the actual structures being analyzed.
For example, the "temporal cortex" could potentially include portions of the nidopallium (including the caudolateral nidopallium, NCL), mesopallium, and arcopallium. Similarly, the "anterior cortex" could correspond to the somatosensory or visual Wulst, the anterior nidopallium, or several other structures. The designation "dorsal cortex" is similarly difficult to interpret. Because these seed regions may encompass multiple functionally distinct systems, it is impossible to evaluate the biological significance of the reported connectivity patterns.
I strongly encourage the authors to define their seed regions using accepted avian neuroanatomical terminology and to provide detailed anatomical maps. More informative analyses would focus on well-defined structures with known connectivity, such as the Wulst, arcopallium, entopallium, or NCL. As currently presented, the tractography results are too coarse to support meaningful biological conclusions.
This is an important concern, and we agree that the anatomical nomenclature and definition of the regions used in the tractography analyses require substantial improvement.
In the revised manuscript, we will carefully re-evaluate the anatomical description of each region with reference to established avian neuroanatomical terminology, anatomical atlases, and available histological information. Where the analyzed regions can be reliably assigned to established anatomical structures, we will replace broad mammalian-style positional terminology such as “anterior cortex,” “dorsal cortex,” and “temporal cortex” with more appropriate avian neuroanatomical terminology. For example, we will re-examine whether the region currently referred to as the “optic lobe” can be more precisely defined as the optic tectum, while the cerebellum can be retained as an anatomically well-defined region.
At the same time, we recognize an important limitation of the present tractography analysis. The spatial resolution and analytical framework of the present comparative datasets do not necessarily permit reliable assignment of all analyzed regions or reconstructed trajectory patterns to fine pallial subdivisions such as the entopallium, arcopallium, or NCL across species. We therefore do not intend to assign such specific anatomical identities where they cannot be supported with sufficient confidence.
Instead, for regions that cannot be unambiguously assigned to a single established anatomical subdivision, we will define their location and extent using reproducible anatomical landmarks and clearly indicate the level of anatomical resolution supported by the data. We will also provide revised anatomical maps showing the locations of the analyzed regions and their relationship to major avian brain subdivisions. This will allow readers to evaluate more clearly which anatomical structures may contribute to the reconstructed trajectory patterns.
Accordingly, we will revise the biological interpretation of the tractography results to match this anatomical resolution. Rather than attributing the reconstructed patterns to specific fine-scale pallial structures or functional systems when these cannot be reliably distinguished, we will interpret them more conservatively as broad patterns of fibre organisation among anatomically defined brain regions. These revisions will improve the anatomical transparency of the analysis while avoiding anatomical or functional interpretations that exceed the resolution of the present datasets.
(3) I am not an MRI specialist, but I have concerns regarding the interpretation of the tractography results. Bird brains are small, and diffusion MRI tractography is already known to be challenging even in substantially larger brains. The manuscript provides limited information regarding image resolution, diffusion sampling, and the expected accuracy of tract reconstruction in these specimens. More importantly, there is little validation of the tractography results. Diffusion tractography is prone to both false positives and false negatives, and reconstructed pathways cannot be assumed to represent true anatomical connections.
The authors should provide evidence that their tractography pipeline can accurately recover known pathways. For example, they could compare reconstructed tracts with well-established anatomical pathways such as the anterior commissure or major visual pathways, which would substantially strengthen confidence in the results. Without such validation, it is difficult to determine whether the observed species differences reflect biological variation or methodological artifacts.
We agree with the reviewer that the tractography results require cautious interpretation and that further evaluation of the robustness and anatomical plausibility of the tractography pipeline would strengthen the study.
In the revised manuscript, we will provide a more detailed description of the diffusion MRI acquisition parameters, spatial and angular resolution, diffusion reconstruction, and tractography procedures so that the methodological limitations of the analyses can be more clearly assessed.
Recent studies using DTI under imaging conditions comparable to those employed in the present study have demonstrated successful reconstruction of fiber trajectories in the mouse brain, which is smaller than the avian brains examined here (Janz et al., eLife, 2017). We therefore consider that the size of the avian brain itself does not preclude DTI-based fiber reconstruction and that such analyses are technically feasible at the brain sizes examined in the present study.
Importantly, we will perform additional sensitivity analyses to evaluate the robustness of the reconstructed trajectory patterns. Specifically, we will examine the effects of varying the FA threshold used for tract reconstruction. We will also re-analyse the available diffusion data using Generalized Q-Sampling Imaging (GQI) as an alternative reconstruction approach and compare the resulting trajectory patterns with those obtained using the current DTI-based analysis. We recognize that the ability to resolve complex fiber configurations depends on the diffusion acquisition parameters, including b-value and spatial and angular resolution. We will therefore interpret the comparison between reconstruction approaches within the limitations of the currently available datasets, without implying that either approach provides a definitive reconstruction of the underlying fiber architecture.
We will also evaluate the robustness of the k-means clustering used to summarize tractography patterns. Because the current choice of k = 10 was not based on an independently established biological criterion, we will examine alternative values of k and assess whether the major trajectory patterns and cross-species differences are robust to the choice of cluster number.
In addition, we will examine anatomically well-characterized pathways, including major commissural and visual pathways, and assess whether the reconstructed trajectories are consistent with known avian neuroanatomy. We will use these comparisons as an assessment of anatomical plausibility rather than as definitive validation of tractography accuracy.
Finally, we will revise the manuscript to more clearly distinguish diffusion MRI-derived reconstructed trajectories from anatomically demonstrated neuronal connections. We will avoid interpreting reconstructed streamlines as direct evidence of anatomical connectivity and will explicitly discuss the possibility of false-positive and false-negative tract reconstruction, as well as other limitations inherent to diffusion MRI tractography.
(4) I also have some methodological concerns regarding the comparisons of anterior commissure (AC) size and cerebellar foliation. First, the authors measure the AC in a coronal section. I would recommend measuring the AC area in a midsagittal section instead. Furthermore, the authors use the cross-sectional area of the same coronal section as the scaling variable. This seems problematic because the area of any given section will depend on the angle of sectioning and other technical factors. If the objective is to compare the relative size of the AC, then total brain volume or telencephalon volume would be more appropriate scaling variables.
With respect to cerebellar foliation, the authors developed their own metric. I would encourage them to use methods already established in the literature, such as the foliation index described by Iwaniuk et al. (2006). Their approach may yield similar results, but using the foliation index would facilitate direct comparisons with existing datasets and would allow incorporation of additional published data (e.g., Cunha et al., 2021, which includes foliation index measurements for 54 bird species). The authors should also be aware that the foliation index scales with body size. Consequently, the high degree of foliation observed in penguins may not necessarily indicate cerebellar expansion or increased demands for sensorimotor integration associated with their specialized locomotion. I therefore believe that the conclusions regarding variation in AC size and cerebellar foliation should be re-evaluated after more appropriate analyses are performed.
These methodological suggestions are very helpful. We agree that both the anterior commissure analysis and the assessment of cerebellar foliation should be re-evaluated to enable more anatomically and quantitatively appropriate comparisons across species.
Anterior commissure:
We agree that the current analysis, in which AC area was measured from the coronal section showing the largest cross-sectional profile and normalized to whole-brain area in the same section, may be influenced by differences in brain geometry and section orientation among species. In the revised manuscript, we will re-evaluate the method used to quantify AC size, including measurements from sagittal or midsagittal views where anatomically appropriate. We will also examine normalization against volumetric measures, such as total brain or telencephalic volume, rather than relying solely on a single-section whole-brain area. The corresponding results and interpretations will be revised accordingly.
Cerebellar foliation:
We also agree that our current branch-counting approach should be considered in relation to established quantitative measures of avian cerebellar foliation. In the revised manuscript, we will evaluate whether the cerebellar foliation index described by Iwaniuk et al. (2006), or a comparable standardized measure applicable to our midsagittal MRI datasets, can be used to re-analyze cerebellar foliation. This will also allow us to place our observations more directly in the context of the larger comparative datasets reported previously, including that of Cunha et al. (2021).
Importantly, we recognize that cerebellar foliation is strongly influenced by allometric and phylogenetic factors. We will therefore re-evaluate the interpretation of interspecific differences in cerebellar foliation in light of brain/body size relationships and the existing comparative literature. In particular, we will revise the interpretation of the pronounced cerebellar foliation observed in penguins and other species and avoid attributing these differences directly to locomotor or sensorimotor specialization unless supported by the revised analyses.
