Introduction

Species of the order Carnivora are a diverse group of mammals comprising at least 270 different species. Carnivorans have colonised various ecological niches, ranging from cursorial and arboreal terrestrial to semi-aquatic and aquatic. They show a range of approaches when interacting with their environments, for example, preferentially using their paws or snouts. Despite their common association with meat eating, carnivorans have a variety of diets, including the completely herbivorous giant panda (Wilson and Mittermeier, 2009). They display a wide range of social structures, from solitary to complex hierarchies, and the order contains wild, fully domesticated, and partially feral animals (MacDonald, 1983). This diversity in behaviour and ecology, coupled with their relatively large brains (Smaers et al., 2021), make carnivorans a key study object for comparative and evolutionary neuroscience.

Large-scale comparisons of different species’ brains have often focused on encephalisation, investigating whether larger relative brain size is associated with particular ecological or social factors (Healy, 2021; Jerison, 1985). Within carnivorans, this work has suggested that increases in relative brain sizes are predominantly associated with environmental factors, especially in the context of a cognitive buffer against environmental instability during foraging (Holekamp and Benson-Amram, 2017; Michaud et al., 2022). However, a full understanding of brain diversity and its potential causes and effects requires a focus not only on brain size but also on underlying aspects of brain organization, including regional variation in size, connections, neurotransmitters, and gene expression (Krubitzer and Kaas, 2005; Mars et al., 2021). Within primates, such comparisons have provided a substantially more detailed understanding of how brains can differ within a specific lineage (Hecht et al., 2013; Hori et al., 2020; Roumazeilles et al., 2020). Similarly, within carnivorans, different dog breeds show that regional differences in brain size correlate with their behavioural specializations (Hecht et al., 2019), and comparisons between domesticated and wild canids reveal regional differences in gyrification (Grewal et al., 2020). Furthermore, novel investigations indicate unique gene expressions in Canidae compared to other mammals (Sacco et al., 2023) and suggest a link between neural complexity and cortical folding in Felidae (Nelson et al., 2024). This shows that a comprehensive understanding of brain diversity in carnivorans requires us to study multiple aspects of brain organization.

Here, we present the first in a series of studies exploring the diversity in carnivoran brain organization, focusing on the identification of the major sulci in surface reconstructions of the neocortex. We exploit the fact that neuroimaging techniques, such as magnetic resonance imaging (MRI), now enable us to collect whole-brain data of various tissue properties quickly, affordably, without damage to the specimen, and in a format that is easily shared between researchers (Mars et al., 2014; Thiebaut de Schotten et al., 2019). Neocortical sulci present clear anatomical landmarks that can be unambiguously established, providing a frame of reference for navigating the neocortex of individual species, thus allowing macro-level comparison across species, as has been done for primates (Amiez et al., 2023; Connolly, 1950). Although there are reports of sulcal anatomy in carnivorans, these either focus on individual species (Chengetanai et al., 2020c), a small range of species (Radinsky, 1975a) or a particular neural system (Lyras and Van Der Geer, 2003; Welker and Campos, 1963) studied in a few species.

In the present study, we provide a comprehensive overview of neocortical sulcal neuroanatomy across eighteen Carnivoran species (Figure 1). We discuss the key differences in sulcal configurations across the brain, which cortical regions likely expanded in particular lineages, and how this might relate to each species’ behaviour and ecology.

Figure 1.

Table 1

Results and discussion

We labelled the neocortical sulci of eighteen carnivoran species (see Figure 1) based on reconstructed surfaces and created standardised criteria (“recipes”) to identify each major sulcus. Our study included five Feliformia and thirteen Caniformia species from seven different carnivoran families. Within the sub-order Caniformia, we examined seven Canidae and six Arctoidea species.

Overall, of the carnivorans studied, Canidae brains exhibited the largest number of unique major sulci, while the brown bear brain was the most gyrencephalic, with the deepest folds and many secondary sulci (see Figures 2-3; brains are arranged by descending number of major sulci). The brown bear was also the largest animal in the sample. The brains of the smallest species, the fennec fox and meerkat, were the most lissencephalic, with the sulci having fewer undulations or indentations compared to the other species. A similar trend has also been observed in the sulci of the prefrontal cortex in primates (Amiez et al., 2023, 2019). The meerkat exhibited the smallest number of major sulci but possessed a unique configuration of sulci in the temporal lobe. In the following, we describe each sulcus’ appearance, the recipes on how to identify them, and provide an overview of the most significant differences across species.

Figure 2.

Figure 3.

Identification of major sulci in the carnivoran neocortex

Occipito-temporal region

Pseudo-sylvian (or sylvian) fissure

The pseudo-sylvian sulcus presents as a fissure on the lateral surface, originating at the ventral border of the neocortex. It extends dorsocaudally, displaying a distinct caudal inclination, and was present in all brains studied (Figure 2, yellow). Analogous to the primate sylvian fissure, the insular cortex can be found within the fissure. However, unlike its primate namesake, the carnivoran pseudo-sylvian fissure is centrally located within the temporal lobe, with occipito- and parietotemporal sulci forming concentric or u-shaped arches around it. As the primate and carnivoran temporal lobe are thought to have evolved independently, the primate sylvian and carnivoran pseudo-sylvian fissure should not be considered homologous (Kaas, 2013; Lyras, 2009). To avoid incorrectly attributing homology between structures, we refer to the fissure in carnivorans as the pseudo-sylvian fissure.

In the majority of carnivoran brains, the pseudo-sylvian fissure curved upwards at its caudal end. Exceptions were noted in the red panda and meerkat, where we observed an open u-shaped fissure. The pseudo-sylvian fissure in the red panda was split at the middle section, but prior observations indicate that it can also have a continuous u-shape (England, 1973). Preliminary evidence suggests that the insular cortex of the red panda (Buchanan and Johnson, 2011) is not fully covered by the opened pseudo-sylvian fissure, which might also be the case for meerkats. We did not observe significant variations in length and shape of the pseudo-sylvian fissure within Canidae and Felidae brains, except that it was shortest in the smallest animal, the fennec fox. Compared to Felidae and Canidae, the pseudo-sylvian fissure of the Asian small-clawed otter, Eurasian badger, South American coati, raccoon and brown bear was elongated and extended more dorsally.

Ectosylvian sulcus

The ectosylvian sulcus varies significantly in shape and occurrence across Carnivoran species. In Canidae brains, the ectosylvian sulcus is the first concentric sulcus folding around the pseudo-sylvian fissure (Figure 2, brown). In Felidae, the sulcus is not a continuous arc but is divided into a caudal and rostral ectosylvian sulcus; only in lions, a study reports variable occurrences of a continuous arc (Sakai et al., 2016).

In our sample, we observed a divided ectosylvian sulcus in all Felidae and a continuous arc in all Canidae, except in the dingo, where the caudal end of the ectosylvian sulcus was split into two parts. The meerkat only exhibited a caudal ectosylvian sulcus, and the Asian small-clawed otter, Eurasian badger, South American coati, raccoon, red panda, and brown bear (i.e., all Arctoidea species) brain did not have an ectosylvian sulcus.

Suprasylvian sulcus

The suprasylvian sulcus is present in all carnivorans with only slight variations (Figures 2-3, purple). In Canidae, it represents the second arc curving around the pseudo-sylvian fissure; in all other species, it is the first complete arc.

In the fennec fox and meerkat, the suprasylvian sulcus was smoother, with no radial branches arising from the sulcus as was observed in the other species. In the meerkat, the caudal end of the sulcus was also less developed and did not bend, but we observed a detached indentation where the caudal bend was observed in other species. Together with the caudal ectosylvian sulcus, the suprasylvian sulcus formed the only (incomplete) arc in the meerkat brain. In the brown bear, Asian small-clawed otter, Eurasian badger and South American coati, the sulcus had an inverted u-shaped form.

Ectomarginal (or ectolateral) sulcus

The ectomarginal sulcus is only present in the canid brains and runs between the marginal and suprasylvian sulcus (Figure 2-3, red). In all canids, the sulcus had a curved pattern similar to the suprasylvian and ectosylvian sulci, but it formed only the caudal half of a complete arc.

Marginal (or lateral, intraparietal) sulcus

The marginal sulcus is an extensive longitudinal sulcus that runs anteriorly from the occipital lobe along the dorsal convexity parallel to the midline and merges with the perpendicularly directed ansate sulcus that lies caudal to the postcruciate and cruciate sulcus in the parietal lobe (Figure 2-3, cyan). The occasionally used term “intraparietal” is suggestive of potential homology with the primate intraparietal sulcus of the same name, but it should be noted that the carnivoran parietal cortex is significantly smaller than that of the primate (Garin et al., 2022; Manger et al., 2002), and thus, the carnivoran sulcus’ territory encompasses other anatomical areas.

The curvature of the marginal sulcus and the length of the caudal portion varied across species. In the meerkat and fennec fox, the sulcus was the relatively smallest. In all other species, the caudal end of the marginal sulcus was more elongated, curving parallel to the suprasylvian, or ectomarginal in Canidae, sulcus. In the brown bear, Asian small-clawed otter, and Eurasian badger, the caudal portion of the marginal sulcus extended even more ventrally than in the other species and almost formed an additional incomplete arc with the ansate and coronal sulcus. In the brown bear, the marginal sulcus extended the most ventral and terminated closely to the caudal rhinal fissure (see Supplementary Figure S2 for ventral view). In several species, the caudal portion of the marginal sulcus was detached, which is a frequently reported occurrence (England, 1973; Kawamura, 1971; Kawamura and Naito, 1978).

Potentially due to the similar caudal bend, some authors label the caudal portion of the marginal sulcus in Ursidae as the ectomarginal sulcus (Lyras et al., 2023, but see e.g., Sienkiewicz et al., 2019); however, unlike the ectomarginal sulcus of Canidae, which runs between the suprasylvian and marginal sulcus, it is a caudal extension of the dorsal part of the Ursidae marginal sulcus.

Endomarginal (or parietal, entolateral, endolateral)

The endomarginal sulcus runs between the marginal sulcus and the median longitudinal fissure on the dorsal surface of the carnivoran brain (Figure 3, yellow).

Although, the endomarginal sulcus has been reported to be present in larger-sized wolf-like canids (Lyras, 2009; Radinsky, 1969), in the present study, it was only identified in domestic dogs and African wild dogs. Prior observations in domestic dogs confirm that the endomarginal sulcus is not always readily identifiable (Czeibert et al., 2018). The brown bear and South American coati had the longest endomarginal sulcus, which is occasionally also referred to as “parietal sulcus” in the brown bear (Sienkiewicz et al., 2019). In the Asian small-clawed otter, the endomarginal sulcus was split into two parts, but as shown previously, it can also appear as a continuous sulcus (Radinsky, 1968). The meerkat had a continuous endomarginal sulcus, which has also been termed the posterior marginal sulcus (Radinsky, 1975a), but due to its position and shape in the meerkat brain, we refer to it as the endomarginal sulcus. The endomarginal sulcus of the Eurasian badger presented as a small groove; the raccoon, red panda and felids did not have an endomarginal sulcus.

Fronto-parietal region

Ansate and coronal sulcus

The rostral end of the marginal sulcus coincides with a complexly organized sulcal region comprising two distinct sulci. The first sulcus runs dorsomedially towards the medial longitudinal fissure and is named the ansate sulcus (Figures 2-3, green). The second sulcus continues rostrally on the lateral surface, curving around the cruciate sulcus, and is termed the coronal sulcus (Figures 2-3, dark blue).

In most species, the coronal sulcus merged with the ansate and marginal sulcus, but in the felids and the meerkat, these sulci did not merge. The coronal sulcus of all Arctoidea species extended more laterally; in the brown bear, Asian small-clawed otter, and red panda, it was also distinctly more elongated compared to the other species.

The racoon exhibited the most elongated ansate sulcus, and in the Asian small-clawed otter, it was the most complex, with multiple perpendicular branches extending in the region between the postcruciate and coronal sulcus. Based on its position and shape, the detached rostral portion of the ansate sulcus in the Asian small-clawed otter resembles the triradiate sulcus of raccoons (Figure 2, light pink), but there are also reports of a continuous sulcus (Welker and Campos, 1963). Conversely, in the fennec fox, bush dog, red panda, meerkat, and European wolf (Figure 2, Supplementary Figure S4B), the ansate sulcus was more difficult to identify, lacking a clear perpendicular form. Here, we labelled the dorsal bend at the rostral end of the marginal sulcus as the ansate sulcus.

Cruciate, postcruciate and triradiate sulcus

The cruciate sulcus (Figures 2-3, light blue) is positioned roughly orthogonal to the median longitudinal fissure on the dorsal convexity of the brain. On the medial wall, it often merges with the dorsal longitudinal splenial sulcus (see Supplementary Figure S3). The postcruciate sulcus (Figure 2-3, pink) is located between the coronal/ansate and the cruciate sulcus. The triradiate sulcus (Figure 2-3, light pink) is a raccoon specialization (Welker and Campos, 1963; Welker and Seidenstein, 1959) situated between the ansate, postcruciate and coronal sulcus.

The cruciate sulcus of the Felidae species examined extended orthogonally from the longitudinal fissure and stayed mostly on the dorsal surface of the brain and was, therefore, not clearly visible from a lateral perspective (Figure 2). In all other species, the sulcus originated at the midline and exhibited a ventrorostral orientation with bends of varying degrees. The cruciate sulcus of the brown bear, Asian-small clawed otter, Eurasian badger, raccoon, and South American coati was the most elongated and curved ventrally. The brown bear also had a short secondary sulcus caudal to the cruciate sulcus, whereas the fennec fox exhibited the least expressed and shortest cruciate sulcus.

Together with the red panda, the species with the most elongated cruciate sulcus (brown bear, Asian-small clawed otter, Eurasian badger, raccoon, and South American coati) also had an extended postcruciate sulcus coupled with a more lateral coronal sulcus (as described above). Due to its complex appearance, the postcruciate sulcus of the red panda and some Procyonidae species, such as the South American coati and raccoon, is often termed the “postcruciate complex” (Welker and Campos, 1963; Welker and Seidenstein, 1959). The Asian small-clawed otter and raccoon had a dorsal postcruciate sulcus running parallel to the midline with a more ventral portion that merged with the perpendicular coronal sulcus. In these cases, the two parts terminated close to each other, but, as observed previously, they occasionally merge (Welker and Campos, 1963; Welker and Seidenstein, 1959). In canids and felids, the postcruciate sulcus was primarily a shallow groove and, therefore, not clearly identifiable in all species. For example, in the bush dog and fennec fox, we noted the postcruciate sulcus only in the left hemisphere and for the domestic dog only in the canine brain template (see Supplementary Figure S4A). Of the Felidae and Canidae, the American wolf and amur leopard had the most expressed postcruciate sulcus; in the latter, it merged with a secondary branch of the ansate sulcus. The meerkat did not have a postcruciate sulcus.

Presylvian sulcus

Rostral to the pseudo-sylvian fissure, the perisylvian sulcus originates from or close to the rostral lateral rhinal fissure (see Supplementary Figure S2 for ventral view). The sulcus extends dorsally, and we observed a gentle caudal curve in the majority of the species (Figures 2-3, white).

There were no major variations across species, but we noted a shortened sulcus in the meerkat and the presence of a secondary branch at the dorsal end that extended rostrally in the Eurasian badger and South American coati brain. The brown bear exhibited an additional sulcus in the frontal lobe, previously labelled as the proreal sulcus (see, e.g., Sienkiewicz et al., 2019); however, the shape of this sulcus in the brown bear closely resembled the secondary branches of the perisylvian sulcus seen in the South American coati and Eurasian badger. Given the known gyrencephaly of Ursidae brains with frequent secondary and tertiary sulci (Lyras et al., 2023), we propose that this sulcus represents a detached branch of the perisylvian sulcus.

Proreal sulcus

The proreal sulcus is the most rostral sulcus; it is located in the frontal lobe and has an axis that runs orthogonal to the presylvian sulcus (Figures 2-3, orange).

This sulcus was present in all canids except for the red and fennec fox (i.e., fox-like canids). We could only identify the proreal sulcus in the dorsal view of the Belgian shepherd brain, but in the group-averaged domestic dog template, it was clearly visible in both dorsal and lateral views (see Supplementary Figure S4C). The sulcus appeared shallow and was, again, only clearly visible from a dorsal perspective in the bush dog (Figure 3).

Sulcal anatomy variations and corresponding gyral differences

Canidae were the only species where an ectomarginal sulcus could be identified, resulting in the formation of the ectomarginal gyrus situated between the marginal and ectomarginal sulcus (Figure 4, dark blue). Additionally, both canids and felids had a (sometimes incomplete) ectosylvian sulcus, defining the outer boundary of the sylvian gyrus, which curves around the pseudo-sylvian fissure (Figure 4, yellow). Consequently, species without an ectosylvian sulcus (i.e., Arctoidea) also lacked a sylvian gyrus. Intuitively, one would assume that they lacked the ectosylvian gyrus; however, in the ferret, which lacks an ectosylvian sulcus, the first gyrus curving around the pseudo-sylvian fissure is consistently referred to as ectosylvian gyrus (see e.g., Radtke-Schuller, 2018). This terminology might be attributed to the cortex forming this gyrus housing the primary auditory cortex, a feature shared with the ectosylvian gyrus of the domestic cat and dog or African wild dog (Bizley et al., 2005; Chengetanai et al., 2020a; Kosmal, 2000; Stolzberg et al., 2017, and see next section). In the meerkat, the caudal ectosylvian sulcus and suprasylvian sulcus form an incomplete arc, which defines the outer boundary of a gyrus curving around the pseudo-sylvian fissure. The suprasylvian sulcus extends further caudal to the incomplete arc and defines the dorsal border of an additional gyrus caudal to the caudal ectosylvian sulcus. To determine if these gyri should be labelled sylvian or ectosylvian, investigations into the location of the primary auditory cortex in Herpestidae are required.

Figure 4.

The presence of the proreal sulcus in wolf-like canids and the greater relative complexity and size of the postcruciate sulcus in the red panda, South American coati, raccoon, Eurasian badger, Asian-small clawed otter, and brown bear likely coincide with an expansion of the proreal gyrus in Canidae and of the postcruciate (or anterior sigmoid gyrus) in Arctoidea (Figure 4, orange, rosé).

Lineage-specific observations and potential relationship to function

The understanding of neocortical areas in the carnivoran brain, such as the location of specific sensory regions, remains incomplete or limited in many species. However, by using the sulcal patterns as a framework and combining them with the knowledge we have about sensory regions in some of the species, we can make inferences about the possible expansion of specific sensory regions and form predictions about their location in less frequent or previously unstudied animals. Furthermore, relating the relative complexity of sulcal topology to the animals’ behavioural and social ecology provides clues regarding potential drivers of neuroanatomical diversity across species.

Somatosensory cortex

One significant variation observed across the species studied was the configuration of the postcruciate and cruciate sulci (see Figure 5A). Compared to all other species, the Asian small-clawed otter, racoon, South American coati, red panda, and brown bear had an expanded postcruciate sulcus with secondary branches and an elongated cruciate sulcus. The Eurasian badger had an elongated cruciate sulcus, while the raccoon possessed an additional sulcus, the triradiate sulcus.

Figure 5.

Prior electrophysiological recordings in the domestic cat (Dykes et al., 1980), domestic dog (Pinto Hamuy et al., 1956), raccoon, coati, and red panda (Welker and Campos, 1963; Welker and Seidenstein, 1959) have revealed that the neocortex surrounding the postcruciate sulcus, the postcruciate (or posterior sigmoid) gyrus, is the location of the primary somatosensory cortex (S1; see Figure 6 for schematic drawings). S1 extends ventrally to the rostral suprasylvian sulcus, covering the rostral suprasylvian (or coronal) gyrus, primarily involved in processing tactile information from the head.

Figure 6.

In the raccoon, red panda, and coati, considerably larger portions of the postcruciate gyrus S1 area appeared to be allocated to representing the forepaw and forelimbs (Welker and Campos, 1963; Welker and Seidenstein, 1959) when compared to the domestic cat or dog (Dykes et al., 1980; Pinto Hamuy et al., 1956). This aligns with the observation that all species in the present sample with more complex postcruciate and cruciate sulci configurations display a preference for using their forepaws when manipulating their environment (see e.g., Iwaniuk et al., 1999; Iwaniuk and Whishaw, 1999; Radinsky, 1968; and Figure 5A). This is suggestive of a potential link between sulcal morphology and a behavioural specialization in Arctoidea (see Radinsky, 1968 for similar observations in otter species).

Occipito-temporal cortical territories

We also observed significant variations in the sulcal topology of the occipitotemporal cortex across species (see Figure 5B). In the Caniformia, the Arctoidea species presented with the relatively least complex sulcal topology in the temporal cortex, with only a single sulcal arc, the suprasylvian sulcus (Figure 6), being noted. In contrast, all canids exhibited the most complex sulcal configuration featuring a second complete arc, the ectosylvian sulcus, in the temporal cortex and the additional ectomarginal sulcus in the occipital cortex. All felids had an ectosylvian sulcus but lacked the middle part present in Canidae and, therefore, only exhibited an incomplete second arc. Studies of digital endocasts in Canidae and Felidae observed the occipito-temporal cortex expansion taking place early in canid brain evolution and reported no observations of an ectomarginal sulcus in felids (Lyras, 2009; Radinsky, 1969). The meerkat had a unique sulcal configuration compared to all other species, with relatively less prominent suprasylvian and ectosylvian sulci forming an incomplete arc.

Previous histological and electrophysiological investigations in the domestic cat (Reale and Imig, 1980), domestic dog (Kosmal, 2000; Tunturi, 1950), and African wild dog (Chengetanai et al., 2020a) indicate that large parts of the cortical territory between the middle and caudal parts of the suprasylvian and ectosylvian sulcus (i.e., on the ectosylvian gyrus) house primary and higher-order auditory regions (see also Figure 6). In the domestic cat, which has a split ectosylvian sulcus, the primary auditory cortex extends ventrally to occupy the cortical area between the sections of the split ectosylvian sulcus (Reale and Imig, 1980). It has been previously suggested that the expansion of the auditory cortex might have caused the split of the ectosylvian sulcus in Felidae brain evolution (Radinsky, 1969). Sensory regions beyond the somatosensory cortex remain unstudied in the species without an ectosylvian sulcus in our sample; however, investigations in the ferret (Mustela putorius furo), a species that lacks an ectosylvian sulcus, showed that the auditory cortex is located ventral to the mid suprasylvian sulcus (Bizley et al., 2005). Thus, the suprasylvian sulcus appears to be a reliable marker of the location of unimodal auditory cortex in Carnivorans.

Moreover, research on the domestic cat (Reale and Imig, 1980; Stolzberg et al., 2017) and domestic dog (Kosmal, 2000; Tunturi, 1950) indicates that the sylvian gyrus, which is only present in the species exhibiting an ectosylvian sulcus (see Figure 4), houses further higher-order auditory, but also visual and multisensory brain regions (see Figure 6 for an overview of unimodal cortical regions). In the domestic dog, multisensory or visual cortical regions expand across the territory between the rostral ectosylvian sulcus and the pseudo-sylvian fissure (Kosmal, 2000; Kosmal et al., 2004), and in the domestic cat along the rostral ectosylvian sulcus (Meredith et al., 2018; Stolzberg et al., 2017). Furthermore, in both species, secondary somatosensory regions are bordering the rostral ectosylvian sulcus area. Thus, the rostral sylvian and ectosylvian cortical regions may represent a multisensory integration hub in felids and canids.

Another unique aspect of canid sulcal anatomy was the ectomarginal sulcus in the occipital lobe. Histological research on the African wild dog (Chengetanai et al., 2020b) and neuroimaging and diffusion MRI in domestic dogs (Andrews et al., 2022; Boch et al., 2021) shows that the ectomarginal gyrus and adjacent sulcal region is part of the extrastriate visual cortex with the territory between the neighbouring marginal sulcus and dorsal convexity housing the primary visual cortex (V1; see Figure 6). The emergence of an ectomarginal sulcus may, therefore, indicate an expansion of the visual cortex in Canidae.

Frontal cortical territory

A distinct proreal sulcus was observed in the frontal lobe of the domestic dog, the African wild dog, wolf, dingo, and bush dog. This may indicate an expansion of frontal cortex in these animals compared to the other species in our sample (Figure 5-6). This aligns with findings from a comprehensive study comparing canid endocasts revealing an expanded proreal gyrus in these animals compared to the fennec fox, red fox and other species of the genus Vulpes (Lyras and Van Der Geer, 2003). The canids with a proreal sulcus also exhibit complex social structures compared to the primarily living solitary foxes (Nowak, 2005; Wilson and Mittermeier, 2009; Wilson, 2000, and see Figure 5). Moreover, a previous investigation of Canidae and Felidae brain evolution, using endocasts of extant and extinct species, also suggested a link between the emergence of pack structures and the proreal sulcus in Canidae (Radinsky, 1969). Despite being highly social and living in large social groups (i.e., mobs), meerkats appear to have a relatively small frontal lobe and no proreal sulcus compared to the social Canids (Figure 5), which would suggest that if the presence of a proreal sulcus correlates with complex social behaviour, this is canid specific.

General discussion

Carnivorans represent a diverse order of mammals, characterised by a range of foraging and social behaviours, possessing relatively encephalised brains. In the present study, we explored cortical sulcal anatomy across a wide range of carnivoran species, marking the first in a series of communications dedicated to exploring carnivoran brain organization. We provide a comprehensive overview of all major lateral and dorsal neocortical sulci in eighteen carnivoran species and standardised recipes to identify each sulcus as a first reference frame to navigate and compare carnivoran brains. The recipes, derived from prior, partially incomplete descriptions and our own observations, provide a unified sulcal nomenclature and were designed to guide future explorations of lesser-studied carnivoran brains. We then conducted a macro-level comparison of observed sulcal configurations across carnivoran species and families and found variations across lineages and species that may relate to how these animals interact with their environment for social structures and forage. Thus, our findings not only establish a framework to guide investigations into carnivoran brain organisation but also generate a large set of potential avenues to be addressed in future investigations.

Comparative analyses of encephalisation provide important insights into the evolutionary trajectory of carnivoran brains and potential factors influencing changes in relative brain size. Prior investigations found, for example, varying encephalisation shifts occurring independently in several families and at different rates (Finarelli and Flynn, 2009; Michaud et al., 2022). These changes in relative brain size were strongly correlated with home range size and geographic distribution in terrestrial Carnivorans (Michaud et al., 2022). However, although encephalisation is a useful measure to compare large numbers of species, its lack of specificity in terms of local expansions and other measures of brain organisation limits its ultimate use. By comparing sulcal anatomy, we observed lineage-specific regional variations in sulcal configuration potentially linked to the animal’s behaviour and ecology, which cannot be detected by comparing relative brain sizes. Comparisons of regional brain size in a sample of social and solitary felids found, for example, a relationship between rostral cerebrum size and sociality in big cats (Sakai et al., 2016). Our work introduces another level, moving from whole-brain brain measurements to investigating patterns of sulcal configuration and potential functional correlates.

Our results revealed several interesting patterns of local variation in sulcal morphology between and within different lineages. For example, Arctoidea showed relatively complex sulcal anatomy in the somatosensory cortex but low complexity in the occipito-temporal regions. In Canidae and Felidae, we found more complex occipito-temporal sulcal patterns indicative of changes in the amount of cortex devoted to visual and auditory processing in these regions. These observations may be linked to social or ecological factors, such as how the animals interact with objects or each other and their varied foraging strategies. Another example was the differential relative expansion of the neocortex surrounding the cruciate sulcus, which was particularly complex in Arctoidea species that are known to use their paws to manipulate their environment. Although it has been argued that the cruciate sulcus appeared independently in different lineages and its exact relationship to the location of primary motor areas varies (Radinsky, 1971), our results provide a detailed exploration of the relationship between brain morphology and behavioural preferences across such a range of species.

Exploring the relationship between relative brain size and sulcal configuration, our findings showed that larger brains were more convoluted, but they did not consistently have the highest number of unique sulci. Despite not having the largest brains (Michaud et al., 2022), Canids exhibited the highest number of unique sulci, and the brown bear with the largest brain in our sample had the most convoluted brain with the deepest folds and numerous secondary sulci. Similarly, while the meerkat, among the smallest species in our study, exhibited a relatively smooth brain surface, its temporal cortex had a unique sulcal configuration. This included a caudal ectosylvian sulcus, which the meerkat shared exclusively with Felidae and Canidae, but the sulcus was absent in Arctoidea species, including the brown bear. These findings underscore that relative brain size alone does not account for the regional variations observed in carnivoran sulcal morphology.

We focused in this work on cortical surfaces reconstructed from whole-brain MRI scans. This contrasts with the more traditional approach of labelling photographs of the samples. There are a number of reasons why labelling MRI scans can be advantageous. The first is that it allows a direct integration with other imaging modalities acquired from the same brains. For example, most of the data reported here were derived from diffusion MRI scans, which will allow us to perform reconstructions of the major white matter bundles in these species (e.g., Jacqmot et al., 2013). Second, the digital nature of MRI data allows more advanced, quantitative comparisons, both within and across species (cf. Mars et al., 2021). For instance, reconstruction of major white matter bundles based on diffusion MRI data has been shown to be a useful measure of comparison across species in primate research (cf. Mars et al., 2018; Warrington et al., 2022) and comparison across different modalities has been shown to be an insightful way to distinguish competing hypotheses of evolutionary change across lineages, such as whether differences are due to local expansion of homologous areas or changes in the connectivity of areas (Eichert et al., 2020, 2019). Third, neuroimaging data are not limited to 2D representations and can be easily shared between researchers, allowing data of rare species and samples to be studied by more groups and using distinct approaches. We have created the Digital Brain Zoo (Tendler et al. 2020) for this specific purpose and the data underlying this project is deposited there.

The sulcal anatomy provided here will equip future work with a reliable reference frame, and the observed potential functional correlates with sulcal morphology provide distinct hypotheses to test. Specifically, the sulcal morphology diversity in the occipitotemporal cortex is intriguing, in particular with respect to the known anatomical variation of the temporal cortex in primates (Braunsdorf et al., 2021), the homology of which across orders remains to be established (Bryant and Preuss, 2018). Formal comparisons in local brain organization can also aid in the further integration of information across levels. For instance, it is known that the dog temporal lobe houses a number of areas specific to social information processing (Andics and Miklósi, 2018; Bálint et al., 2023; Boch et al., 2024, 2023b, 2023a). This dovetails with our observation of the complex sulcal organization of this part of the brain in more social species (Figure 5B). Previous comparative work on the carnivoran social brain has highlighted the extended proreal cortex in social species (Holekamp et al., 2007; Radinsky, 1969). This is a finding that we replicate in the form of the presence of a proreal sulcus in these animals, but it also shows how focusing purely on measures of brain encephalisation reveals only a limited picture. A similar observation has been made in primates, where larger brains tend to correlate with larger social groups (Dunbar and Shultz, 2007), but this finding does not describe the complexity of the homology of brain regions processing social information across species (Mars et al., 2013; Roumazeilles et al., 2021; Wittmann et al., 2018). As another case in point, the meerkat, a highly social species, does not have a proreal sulcus but does have an expanded proreal gyrus compared to the less social mongoose species (Radinsky, 1975a).

In summary, our study transcends traditional brain size comparisons, illustrating the diversity of carnivoran brain organization and regional differences. We propose potential links between the observed variation in sulcal anatomy and the species’ behavioural ecology, and by establishing standardized criteria, we lay the groundwork for a better understanding of lesser-studied carnivoran brains. Finally, we provide the roadmap for further anatomical investigations exploiting the use of neuroimaging in the comparative study of carnivoran and mammalian brain diversity.

Limitations and future directions

Our findings represent a critical first step for linking brains within and across species for interspecies insights. The present analyses are based on multiple individuals pooled into families and genera, focusing on a single representative per species. Future studies will aim at investigating interindividual variability and will extend to more detailed investigations of the medial part of the cortex, as well as the subcortical structures and the cerebellum.

Materials and methods

We obtained the diverse carnivoran brain data mainly through post-mortem samples. Due to variations in scanning protocols (Table 1), we homogenized the data set during the initial preprocessing step. Following that, we employed a standardized pipeline to generate cortical surfaces for all brains. We then used the neocortical surface reconstructions to systematically label all major neocortical sulci in each brain. To guide the labelling process, we created standardized criteria (“recipes”) containing detailed descriptions of how to identify each sulcus based on available prior descriptions and our own observations.

Data

The sample consists of eighteen carnivoran species (Figure 1). All data, except for the cat, were acquired ex-vivo (see Table 1 for detailed sample descriptives). Analyses reported in the main text are all based on a single, adult or subadult individual of each species. We show confirmatory analyses on additional samples or averaged templates in the supplementary material where available.

Procedure

Brains of all post-mortem samples were extracted and processed within 24 hours of death. The majority of the samples were obtained by the Copenhagen Zoo and the Zoological Society of London (see Table 1 for details). The Copenhagen Zoo specimens were all collected immediately after euthanasia. Following general anaesthesia and an overdose of sodium pentobarbital (200 mg/kg, i.v.), the heads of animals were perfusion-fixed post-mortem through the carotid arteries, initially with a rinse of 0.9% saline followed by 4% paraformaldehyde in 0.1M phosphate buffer. The brains were then removed from the skull and post-fixed in paraformaldehyde (PFA) for 24 hours and subsequently transferred to a sodium azide phosphate buffer solution to ensure preservation until MRI data acquisition. Antemortem observations for all these animals revealed that they were in good health with no obvious neural deficits or behavioural abnormalities but were euthanised for population management reasons (Bertelsen, 2019). Samples from the Zoological Society of London’s pathology archive were all from captive animals, with the exception of the Eurasian badger, which was a wild animal found dead on site. The brains had been removed as part of routine post-mortem examinations and fixed and stored in 10% neutral buffered formalin. No evidence of brain pathology was noticed during the necropsy.

The raccoon was euthanised as part of pest control measures by the city of Iwamizawa (Hokkaido, Japan), a decision unrelated to the current study. After extraction, the raccoon brain was immersed in 4% PFA for five days and subsequently transferred to a sodium azide phosphate buffer solution to ensure preservation until MRI data acquisition.

The fennec and red fox brain specimens were removed within 14 hr of death and immersion fixed in 10% formalin at necropsy before being transferred to a solution of 0.1 M phosphate-buffered saline (PBS) with 0.1% sodium azide solution and stored at 4°C prior to scanning. The fennec fox specimen was raised in captivity at the St. Louis Zoo before being donated to M.A.S., while the red fox was donated to M.A.S. by a local taxidermist.

To obtain in-vivo data of the domestic cat, a board-certified anaesthesiologist performed general anaesthesia on the animal. It was premedicated with dexmedetomidine (10mg/kg; Zoetis Inc, Kalamazoo, MI), induced using propofol for general anaesthesia and dosed to effect (10-20 mg/kg; Sagent Pharmaceuticals, Schaumburg, Ill), and intubated. The animal was maintained under anesthesia using oxygen and inhalant isoflurane and supportive intravenous lactated ringer’s solution fluids.

Ethics statement

The in-vivo data collection of the domestic cat was approved by the institutional animal care and use committee at Cornell University (protocol number 2015-0115); and a board-certified veterinary anaesthesiologist (American College of Veterinary Anesthesia) approved all protocols. As outlined above, all ex-vivo data was obtained from animals euthanised or deceased for reasons unrelated to this study.

Imaging protocols and data preprocessing

Data was obtained using 3 and 7T wide-bore human MRI scanners or narrow-bore non-human animal scanners (see Table 1). Scanning protocols included post-mortem T2-weighted and in-vivo T1-weighted structural or post-mortem diffusion protocols. In all cases, we preprocessed the data using tools from FSL (www.fmrib.ox.ac.uk/fsl), ANTs (Avants et al., 2009), and custom-code from the Comparative Anatomy Toolbox (Mr Cat; www.neuroecologylab.org) to create scans with a T1-like contrast needed for surface reconstruction.

Large post-mortem samples from the Copenhagen Zoo specimen collection and the Zoological Society of London (see Table 1) were scanned at the University of Oxford on a wide-bore human 7T scanner with 70 mT/m maximum gradient. Diffusion MRI data were acquired using a diffusion-weighted steady-state free-precession (DW-SSFP) sequence (q-value = 300 cm^-1, gradient duration = 13.56 ms, gradient amplitude = 52 mTm^-1, flip angle = 39°, 10 to 13 non-diffusion weighted volumes per brain, TE/TR = 21/29 ms, EPI factor = 1, Bandwidth = 100 Hz per pixel) at 600 μm isotropic resolution with 160 diffusion directions, using a 1 Tx/28 Rx QED knee coil (European wolf, Asiatic lion, brown bear) or 1 Tx/32 Rx Nova head coil (American wolf, amur leopard, African wild dog). The acquisition took 16 min and 25 s per volume. Datasets were corrected for Gibbs ringing (Kellner et al., 2016) and co-registered using FSL FLIRT (Jenkinson and Smith, 2001). Diffusion Tensor and Ball & 2 Sticks model estimates were derived using custom software accounting for the full DW-SSFP signal model (Buxton, 1993), implemented using cuDIMOT (Hernandez-Fernandez et al., 2019). To account for the dependencies of DW-SSFP on relaxation times and flip angle, quantitative T₁, T₂ and B₁ maps were additionally estimated in each sample using a turbo inversion-recovery, turbo spin-echo and actual flip angle imaging (AFI) sequence (Yarnykh, 2007), respectively. A ball and two stick model was fitted to the data using a modified version of bedpostX (Behrens et al., 2007). Diffusion modelling code is available at https://github.com/BenjaminTendler/DW-SSFP).

Small post-mortem samples from the Copenhagen Zoo specimen collection, the London Zoological Society, and the domestic dog were scanned at the University of Oxford using a narrow-bore rodent 7T MRI scanner (Varian, Oxford UK) with 400 mT/m maximum gradient. Acquisition used a 2D diffusion-weighted spin-echo multi-slice protocol with single line readout (DW-SEMS; TR/TE = 19/26 ms; matrix size = 128 × 128 with a sufficient number of slices to cover each brain; spatial resolution for the dingo, domestic dog (Belgian shepherd), and lynx was .6 mm isotropic, for the bush dog, red panda and coati: .5 mm isotropic, for the Eurasian badger .4 mm isotropic, and .3 mm for the meerkat. A total of 16 non-diffusion-weighted (b = 0 s/mm²) and 128 diffusion-weighted (b = 4000 s/mm2) volumes were acquired with diffusion encoding directions evenly distributed over the whole sphere (single shell protocol). All data were preprocessed using the same protocol implemented in the module phoenix of the MR Comparative Anatomy Toolbox (Mr Cat; www.neuroecologylab.org). Briefly, the steps are as follows: We first converted the datasets to NIFTI format, then built an image based on the volumes acquired without a diffusion gradient as well as a binary mask of this image. Diffusion Tensor and Ball & 3 Sticks model estimates were derived using dtifit and BedpostX tools from FSL (www.fmrib.ox.ac.uk/fsl).

The structural scan of the raccoon brain was obtained at the Medical School of Hokkaido University using a wide-bore human 3T MRI scanner and a 16-channel body coil. The acquisition employed a variable flip-angle turbo spin echo (3D-SPACE) sequence with 172 slices (slice thickness = .63 mm), .63 mm isotropic spatial resolution, and echo train length = 179. Additional parameters included TR/TE = 4000/411 ms, FoV = 107 x 120 mm, matrix 192 x 172, flip angle = 120°, and a bandwidth of 400 Hz per pixel. The scan lasted approximately 10 minutes and 56 seconds.

The in-vivo structural scan of the domestic cat was performed at Cornell Magnetic Resonance Imaging Facility (CMRIF) using a wide-bore human 3T (General Electric Discovery MR750), with 50 mT/m maximum gradient strength. The animal was placed with the head centred and in dorsal recumbency in a 16-channel small flex radiofrequency coil (Neocoil, Pewaukee, WI © NeoCoil). Data was acquired using a T1-weighted 3D MPRAGE (BRAVO) sequence with .5 mm³ isotropic voxel resolution and TR/TE of 8.436/3.604 (TI = 450, flip angle = 12°, NEX = 1). The scan lasted approximately 6 minutes and 30 seconds. Before surface creation, the scan was intensity bias-corrected using ANTs (Avants et al., 2009).

Scanning of the red and fennec fox was undertaken at the Icahn School of Medicine (Mt. Sinai, NY) using a narrow-bore rodent 7T Bruker Biospec scanner. A 3D FLASH (fast low angle shot) sequence was used with the following parameter settings: TR/TE= 36/23 ms, flip angle = 15°, FOV = 128 × 128 × 175, matrix size = 384 × 384 × 384 mm in each slab with a spatial resolution of .13 mm isotropic for the fennec fox and .18 mm isotropic for the red fox.The scans were then intensity bias-field corrected using ANTs before surface creation.

Surface creation

All scans were reoriented to anterior/posterior commissure and standard FSL orientation. We then generated cortical surface meshes using precon_all (https://github.com/neurabenn/precon_all), an adapted version of Freesurfer’s recon-all pipeline (Fischl, 2012), designed to create surfaces of non-human animal models.

Precon_all reconstructs the surfaces based on scans with a T1-like contrast. We, therefore, inverted the voxel intensities of the T2-weighted scans (red fox, fennec fox, raccoon) by multiplying them with −1 and keeping cerebrospinal fluid (csf) intensities at zero. Precon_all requires a spatial resolution of .2 mm or higher. We, therefore, upsampled the scans to .5 mm³ for the red and .3 mm³ for the (smaller) fennec fox using FSL FLIRT (Jenkinson and Smith, 2001) to strike a balance between preserving information and achieving optimal surface generation. For the diffusion MRI scans, we created the T1-like images by calculating the square root of the sum of the mean_f{1,2,3}_samples (indicating the mean of the anisotropy distribution at each voxel).

Next, we removed any remaining non-brain tissue from all brain scans using individual brain masks, which we created using FSL’s bet or ITK-snap’s segmentation tool (Yushkevich et al., 2006) and, if necessary, manually improved them. We then created masks for each hemisphere using FSLeyes and drew “subcortical” and “noncortical” masks in ITK-snap (see Supplementary Figure S1 for example). The subcortical masks comprised the corpus callosum expanding to the outer borders of the lateral ventricles, which are filled during surface generation. We included the cerebellum, brain stem and olfactory bulb in the noncortical mask, to remove them from the surfaces. Following an initial run of precon_all, surfaces were refined by manually editing the resulting white matter masks. We then ran precon_all again to generate white, pial, and mid-thickness surfaces and down-sampled them to 10,000 vertices; and applied spatial smoothing using connectome workbench tools (Marcus et al., 2011) to facilitate sulcal labelling. For the labelling of lateral, ventral and medial wall sulci, we used a smoothing strength of .5, and to detect partially more shallow dorsal sulci, we labelled surfaces with a .2 smoothing strength (15 iterations each).

Labelling and creation of recipes

We focused on the major lateral and dorsal sulci of the carnivoran brain, but the medial wall and ventral view of the sulci are also described. For consistency, we always labelled the right hemispheres on the mid-thickness surfaces. We aimed to facilitate interspecies comparisons and the exploration of previously undescribed carnivoran brains. To this end, we first created standardized criteria (henceforth referred to as recipes) for identifying each sulcus, drawing from existing literature on carnivoran neuroanatomy, particularly in paleoneurology (Lyras et al., 2023), and our own observations. Anatomical nomenclature primarily follows the recommendations of (Czeibert et al., 2018); if applicable, alternative names of sulci are provided once.

We started creating the recipes based on Felidae and Canidae neuroanatomy descriptions since the outer morphology and evolutionary history of these species’ brains are the most extensively described within the order Carnivora (see e.g., Chengetanai et al., 2020c; Lyras, 2009; Radinsky, 1973a, 1975b, 1969; Rogers Flattery et al., 2023; Sakai et al., 2016). We then completed the recipes with the observations made while labelling the other carnivoran brains and, if available, confirmed them with prior neuroanatomical descriptions. Ursidae brains have also been described more frequently (Sienkiewicz et al., 2019; see e.g., Smith, 1933), but only a few and often partial descriptions existed for the other species (but see England, 1973 for comparisons of a wide variety of species). Prior descriptions of the Asian-small clawed otter, South American coati, raccoon and red panda brain primarily focus on somatosensory and motor cortices (Hardin et al., 1968; Johnson et al., 1982; Radinsky, 1968; Welker and Campos, 1963; Welker and Seidenstein, 1959). While descriptions and schematic drawings of the Eurasian badger and meerkat exist, not all sulci are labelled and described in detail, as the investigations focused on other aspects (Radinsky, 1975a, 1973b). If not mentioned otherwise, our observations were in line with available prior descriptions of these species’ neuroanatomy.

We then briefly illustrated the gyri of the carnivoran brain with a focus on gyri that are not present in some species as a consequence of absent sulci to complement our observations. Finally, we summarised the key differences and similarities between species, related them to their ecology and created schematic drawings of the observed sulcal patterns. If known, we also denoted approximate locations of sensory regions to add another layer of comparison that allows for indications of cortical areas that potentially expanded. We obtained information about sensory properties from prior histological and electrophysiological research. For the majority of the species, functional properties of brain areas have not been studied yet, or the focus was only on the somatosensory cortex, such as for the South American coati and red panda (Welker and Campos, 1963). The sensory cortices of the domestic cat are best understood and studied most extensively (see e.g., Stolzberg et al., 2017 for a detailed atlas), followed by the domestic dog (Adrian, 1941; Kosmal, 2000; Kosmal et al., 2004; Pinto Hamuy et al., 1956; Tunturi, 1944), raccoon (Douglas Jameson et al., 1968; Hardin et al., 1968; Welker and Seidenstein, 1959), and African wild dog (Chengetanai et al., 2020a, 2020b).

Data availability statement

Generated surfaces of all species and T1-like contrast images of post-mortem samples obtained by the Copenhagen Zoo and the Zoological Society of London (see Table 1) are available at the Digital Brain Zoo of the University of Oxford (Tendler et al., 2022) (https://open.win.ox.ac.uk/DigitalBrainBank/#/datasets/zoo).

Acknowledgements

The authors would like to thank Boras Zoo, Parken Zoo, Randers Regnskov and Ree Park for contributing samples to the Copenhagen Zoo specimen collection and João Paulo Coimbra for his help collecting the data. For the purpose of Open Access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission.

Funding

M.B. received support for this project through the Postdoc Award granted by the Faculty of Psychology, University of Vienna. The work of R.B.M. is supported by the Biotechnology and Biological Sciences Research Council (BBSRC) UK [BB/N019814/1]. C.L. acknowledges partial funding for the research described herein from the Austrian Science Fund (FWF) [P34675]. M.A.S. received support from the Verizon Foundation. B.C.T is supported by a Sir Henry Wellcome Postdoctoral Fellowship (Wellcome Trust) [222829/Z/21/Z]. K.L.M is supported by the Wellcome Trust [202788/Z/16/Z, 224573/Z/21/Z]. The Wellcome Centre for Integrative Neuroimaging is supported by core funding from the Wellcome Trust [203139/Z/16/Z]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests

The authors declare no competing interests.