Longitudinal connections and the organization of the temporal cortex in macaques, great apes, and humans

Abstract

The temporal association cortex is considered a primate specialization and is involved in complex behaviors, with some, such as language, particularly characteristic of humans. The emergence of these behaviors has been linked to major differences in temporal lobe white matter in humans compared with monkeys. It is unknown, however, how the organization of the temporal lobe differs across several anthropoid primates. Therefore, we systematically compared the organization of the major temporal lobe white matter tracts in the human, gorilla, and chimpanzee great apes and in the macaque monkey. We show that humans and great apes, in particular the chimpanzee, exhibit an expanded and more complex occipital-temporal white matter system; additionally, in humans, the invasion of dorsal tracts into the temporal lobe provides a further specialization. We demonstrate the reorganization of different tracts along the primate evolutionary tree, including distinctive connectivity of human temporal gray matter.

Fig 1

A 3D representation and coronal sections of longitudinal temporal tracts in macaques (A) and humans (B). Tractograms were log transformed and normalized for display. The coronal sections from left to right are taken from posterior to anterior. Blue, MdLF; green, IFOF; yellow, ILF (macaque) or ILFmed (human); red, ILFlat. Thresholds for the tracts are as follows: 0.7 for MDLF; 0.75 for IFOF; and 0.7 for ILF, ILFmed, and ILFlat. Human data are available from the Human Connectome Project (www.humanconnectome.org). Macaque postmortem data are available from the PRIME-DE repository (http://fcon_1000.projects.nitrc.org/indi/PRIME/oxford2.html) [24]. IFOF, inferior fronto-occipital fascicle; ILF, inferior longitudinal fascicle; ILFlat, inferior longitudinal fascicle lateral; ILFmed, inferior longitudinal fascicle medial; MdLF, middle longitudinal fascicle; PRIME-DE, Primate Data Exchange.

Fig 2

Surface projection of longitudinal temporal tracts in macaques (A) and humans (B). Shown are the group averages of the normalized, log-transformed, and thresholded tracts in the right hemisphere only (left hemisphere results can be found in S2 Fig). Blue, MdLF; green, IFOF; yellow, ILF (macaque) or ILFmed (humans); red, ILFlat. Thresholds for the tracts are as follows: 0.7 for MDLF; 0.75 for IFOF; and 0.7 for ILF, ILFmed, and ILFlat. Human data are available from the Human Connectome Project (www.humanconnectome.org). Macaque postmortem data are available from the PRIME-DE repository (http://fcon_1000.projects.nitrc.org/indi/PRIME/oxford2.html). IFOF, inferior fronto-occipital fascicle; ILF, inferior longitudinal fascicle; ILFlat, inferior longitudinal fascicle lateral; ILFmed, inferior longitudinal fascicle medial; MdLF, middle longitudinal fascicle; PRIME-DE, Primate Data Exchange.

Fig 3

A 3D representation of longitudinal temporal tracts in chimpanzees in vivo (A) postmortem (B) and gorilla (C). Tractograms were log transformed and normalized for display. The coronal sections from left to right are taken from posterior to anterior. Blue, MdLF; green, IFOF; yellow, ILFmed; red, ILFlat. Thresholds for the tracts are as follows: 0.7 for MDLF; 0.75 for IFOF; and 0.7 for ILFmed and ILFlat. In vivo chimpanzee data are available from the National Chimpanzee Brain Resource (www.chimpanzeebrain.org). Gorilla and chimpanzee postmortem data are available from https://doi.org/10.5281/zenodo.3901205. IFOF, inferior fronto-occipital fascicle; ILF, inferior longitudinal fascicle; ILFlat, inferior longitudinal fascicle lateral; ILFmed, inferior longitudinal fascicle medial; MdLF, middle longitudinal fascicle.

Fig 4

Surface representation of longitudinal temporal tracts in chimpanzees (A) and gorilla (B). Shown are the group average (chimpanzee in vivo) or individual results (chimpanzee postmortem and gorilla) of the normalized, log-transformed, smoothed, and thresholded right tracts (left hemisphere results can be found in S3 Fig for chimpanzee in vivo). Blue, MdLF; green, IFOF; yellow, ILFmed; red, ILFlat. Thresholds for the tracts are as follows: 0.7 for MDLF; 0.75 for IFOF; and 0.7 for ILFmed and ILFlat. In vivo chimpanzee data are available from the National Chimpanzee Brain Resource (www.chimpanzeebrain.org). Gorilla and chimpanzee postmortem data are available from https://doi.org/10.5281/zenodo.3901205. IFOF, inferior fronto-occipital fascicle; ILFlat, inferior longitudinal fascicle lateral; ILFmed, inferior longitudinal fascicle medial; MdLF, middle longitudinal fascicle.

Fig 5. AF tractography protocols and results.

Top panel: AF tractography masks example for one individual macaque, human, chimpanzee (in vivo and postmortem), and gorilla, represented on the principal eigenvector (V1) map weighted by the fractional anisotropy map. The light-pink mask represents the seed, and the white masks represent the anterior and posterior waypoints. Bottom panel: 3D and surface representation of the right tractogram obtained for AF. (Left hemisphere results can be found in S4 Fig) Threshold of 0.75. R denotes right hemisphere. Human data are available from the Human Connectome Project (www.humanconnectome.org). Macaque postmortem data are available from the PRIME-DE repository (http://fcon_1000.projects.nitrc.org/indi/PRIME/oxford2.html). In vivo chimpanzee data are available from the National Chimpanzee Brain Resource (www.chimpanzeebrain.org). Gorilla and chimpanzee postmortem data are available from https://doi.org/10.5281/zenodo.3901205. AF, arcuate fascicle; PRIME-DE, Primate Data Exchange.

Fig 6. Results of the blueprint analysis.

The figure reports results on the right hemisphere (left hemisphere results can be found in S6 Fig) and shows the tracts used (top row), the resulting KL divergence (middle row) for human predicting macaque (left) and macaque predicting human (right), and the connectivity fingerprints (bottom row). The greener the vertices of the KL divergence map in one brain, the more their connectivity profile is similar to that of vertices in the other brain. The connectivity fingerprints show how likely it is for the vertex, highlighted by a white sphere in the brain above, to be reached by each tract (dotted line). The solid line represents how likely each tract is to reach the average of the 10 best-matching vertices in the other species. (A) Blueprints established using the cbt, the fx, the or, and the unc. (B) Blueprints established with the tracts in (A) and adding the MdLF, ILF (both subcomponents combined in humans), and IFOF. (C) Blueprint established with the tracts in (B) and adding the AF. Blue, MdLF; green, IFOF; yellow, ILF (macaque) or ILFmed (human); red, ILFlat; pink, AF. (D) Distribution of minimum KL values obtained for each blueprint. From left to right for the macaque, the human, and masked with the human AF. Light gray, nonlongitudinal tracts; gray, adding MdLF, ILFs, and IFOF; black, adding AF. Human data are available from the Human Connectome Project (www.humanconnectome.org). Macaque postmortem data are available from the PRIME-DE repository (http://fcon_1000.projects.nitrc.org/indi/PRIME/oxford2.html). AF, arcuate fascicle; cbt, temporal part of the cingulum bundle; fx, fornix; IFOF, inferior fronto-occipital fascicle; ILF, inferior longitudinal fascicle; ILFlat, inferior longitudinal fascicle lateral; ILFmed, inferior longitudinal fascicle medial; KL, Kullback–Liebler; MdLF, middle longitudinal fascicle; or, optic radiation; unc, uncinate fascicle; PRIME-DE, Primate Data Exchange.

Fig 7. Blueprint method to determine differences between the human ILFs.

(A) Blueprint established as in Fig 6C but with the ILFlat for humans (no ILFmed). (B) Blueprint established as in (A) but with the ILFmed for humans (no ILFlat). (C) The different ILF tractograms are represented on the top row. On the bottom row is shown the KL difference between the two maps established in (A) and (B). More yellow vertices means that using ILFmed in humans as macaques’ ILF homologous results in lower KL divergence between the two species at these vertices, whereas more red applies to ILFlat. Blue, MdLF; green, IFOF; yellow, ILF (macaque) or ILFmed (human); red, ILFlat; pink, AF. (D) Distribution of minimum KL values in the macaque’s ILF territory obtained for the blueprints with the different human ILF subcomponents. Red, with human ILFlat; yellow, with human ILFmed. Human data are available from the Human Connectome Project (www.humanconnectome.org). Macaque postmortem data are available from the PRIME-DE repository (http://fcon_1000.projects.nitrc.org/indi/PRIME/oxford2.html). IFOF, inferior fronto-occipital fascicle; ILF, inferior longitudinal fascicle; ILFlat, inferior longitudinal fascicle lateral; ILFmed, inferior longitudinal fascicle medial; KL, Kullback–Liebler; MdLF, middle longitudinal fascicle; PRIME-DE, Primate Data Exchange.

Fig 8. Suggested evolutionary trajectory for the longitudinal temporal lobe tracts.

In the top panel are represented the major temporal anatomical landmarks in rhesus macaque, human, chimpanzee, and gorilla (brains not to scale). In the bottom panel, we represented the temporal tracts obtained in the different species. Blue, MdLF; green, IFOF; yellow, ILF (macaque) or ILFmed (human); red, ILFlat; pink, AF. Human data are available from the Human Connectome Project (www.humanconnectome.org). Macaque postmortem data are available from the PRIME-DE repository (http://fcon_1000.projects.nitrc.org/indi/PRIME/oxford2.html). In vivo chimpanzee data are available from the National Chimpanzee Brain Resource (www.chimpanzeebrain.org). Gorilla postmortem data are available from https://doi.org/10.5281/zenodo.3901205. AF, arcuate fascicle; FG, fusiform gyrus; IFOF, inferior fronto-occipital fascicle; ILF, inferior longitudinal fascicle; ILFlat, inferior longitudinal fascicle lateral; ILFmed, inferior longitudinal fascicle medial; ITG, inferior temporal gyrus; MdLF, middle longitudinal fascicle; MTG, middle temporal gyrus; PRIME-DE, Primate Data Exchange; STG, superior temporal gyrus.

Fig 9. Methods overview.

The macaque is shown as the example; the same methods were used for the other species, except that postmortem analyses were fully performed in subject space. (A) Connectivity-based clustering: (a) In each hemisphere, we defined three coronal ROIs in the white matter of the temporal lobe (excluding the area of the cingulum bundle). We transformed these ROIs to the diffusion space of each individual. (b) We then performed tractography from all the voxels in the ROI to the rest of the voxels in the whole brain. (c) We then computed the similarity matrix, representing the similarity in whole-brain connectivity of all voxels’ in each ROI. (d) We clustered the similarity matrix using the k-means algorithm, which resulted in clusters in the white matter ROI showing which voxels have similar connectivity to the rest of the brain. (e) These clusters were back projected onto each individual brain and transformed to standard space to give what we refer to as the “clustering results” (detailed in Figs 10 and 11). (B) Probabilistic tractography: from the clusters obtained at the clustering step, we established the masks for the tract reconstruction (detailed in Figs 10 and 11). Probabilistic tractography was performed as follows: streamlines were sent from the seed and were kept only if they went through one of the waypoints. The resulting tractograms were averaged across individuals. IFOF, inferior fronto-occipital fascicle; ILF, inferior longitudinal fascicle; MdLF, middle longitudinal fascicle; ROI, region of interest.

Fig 10

Tractography masks (bottom row) derived from clustering results (top row) in macaques and humans. (A) In macaques, the clustering results are thresholded to show the overlap between at least two subjects out of four. (B) In humans, the clustering results are thresholded to show the overlap between at least four subjects out of 10. Blue, MdLF; green, IFOF; yellow, ILF (macaque) or ILFmed (humans); red, ILFlat. R denotes right hemisphere. Human data are available from the Human Connectome Project (www.humanconnectome.org). Macaque postmortem data are available from the PRIME-DE repository (http://fcon_1000.projects.nitrc.org/indi/PRIME/oxford2.html). IFOF, inferior fronto-occipital fascicle; ILF, inferior longitudinal fascicle; ILFlat, inferior longitudinal fascicle lateral; ILFmed, inferior longitudinal fascicle medial; MdLF, middle longitudinal fascicle; PRIME-DE, Primate Data Exchange.

Fig 11

Tractography masks (bottom row) derived from clustering results (top row) in great apes. In chimpanzee in vivo, the clustering results show the three subjects overlapped (no thresholding is applied). One subject was available for each of the chimpanzee and gorilla postmortem samples. Blue, MdLF; green, IFOF; yellow, ILFmed; red, ILFlat. R denotes right hemisphere. In vivo chimpanzee data are available from the National Chimpanzee Brain Resource (www.chimpanzeebrain.org). Gorilla and chimpanzee postmortem data are available from https://doi.org/10.5281/zenodo.3901205. IFOF, inferior fronto-occipital fascicle; ILF, inferior longitudinal fascicle; ILFlat, inferior longitudinal fascicle lateral; ILFmed, inferior longitudinal fascicle medial; MdLF, middle longitudinal fascicle.

Fig 12. Blueprint method to assess connectivity divergence between macaque and human temporal cortex.

(A) To compute the blueprint, we multiplied a matrix representing the connectivity of each cortical vertex (surface space) to each brain voxel (volume space) with a matrix representing how each brain voxel (volume space) is reached by each tract. Because we reconstructed the same tracts in both macaques and humans, the columns of the blueprint define a common space to compare the connectivity of the temporal cortex. The rows of the blueprint correspond to the connectivity fingerprints of each vertex. The columns of the matrix correspond to the surface tractograms. (B) Using the KL divergence measure to compare every vertex in one brain to every vertex in the other brain, we can identify vertices in the two brains in which the smallest divergence is found. If this smallest divergence is very low (green on the scale), it means that the same connectivity profile can be found at these vertices for both brains. If the smallest divergence is very high (pink on the scale), it means that the connectivity profile at this vertex is very different to any connectivity profile of the other brain vertices. (C) Illustration of the KL divergence computation for two macaque vertices. For each of the vertices, a connectivity fingerprint representing how the vertex is connected to each tract was computed. Using a KL divergence measure, this connectivity fingerprint was compared with connectivity fingerprints of all human vertices. The vertex with the connectivity fingerprint most similar to the macaque one, therefore with the smallest KL divergence, is picked, and this KL divergence value is assigned to the monkey vertex. AF, arcuate fascicle; cbt, temporal part of the cingulum bundle; fx, fornix; IFOF, inferior fronto-occipital fascicle; ILF, inferior longitudinal fascicle; KL, Kullback–Liebler; MdLF, middle longitudinal fascicle; or, optic radiation; unc, uncinate fascicle.