The brain of the North American cheetah-like cat Miracinonyx trumani

Summary

The cheetah Acinonyx jubatus, the fastest living land mammal, is an atypical member of the family Felidae. The extinct feline Miracinonyx trumani, known as the North American cheetah, is thought to have convergently evolved with Acinonyx to pursue fast and open-country prey across prairies and steppe environments of the North American Pleistocene. The brain of Acinonyx is unique among the living felids, but it is unknown whether the brain of the extinct M. trumani is convergent to that of Acinonyx. Here, we investigate the brain of M. trumani from a cranium endocast, using a comparative sample of other big cats. We demonstrate that the brain of M. trumani was different from that of the living A. jubatus. Indeed, its brain shows a unique combination of traits among living cats. This suggests that the case of extreme convergence between Miracinonyx and its living Old World vicar should be reconsidered.

Graphical abstract

Highlights

• •The brain of Miracinonyx exhibits a unique pattern of gyri and sulci
• •The anterior cerebrum of Miracinonyx is not as reduced as the one of the cheetah
• •The brain shape of Miracinonyx is not as dorsiflexed as the one of the cheetah
• •The brains of the cheetah and Miracinonyx are not as convergent as expected

Zoology; Evolutionary biology; Paleobiology

Introduction

The living cheetah (Acinonyx jubatus) is widely acknowledged as the fastest living land mammal capable of speeds up to 25.9 ms⁻¹ and reaching top accelerations in only 3 s.¹ Most felids are stalking predators in closed habitat that rely on stealth to approach their prey, followed by a brief, high-speed pursuit.² Although they can accelerate rapidly, they tire quickly and their attacks are often aborted when prey detects them during ambush or early in their approach.³ The only exception to this hunting behavior among the living felids is the cheetah, a diurnal, coursing predator with a highly specialized anatomy and physiology that relies on rapid acceleration and sprinting^4,5,6 over chases of some hundred meters (average run distance: 173 ± 116 m¹). In contrast to stalkers, the cheetah approaches its prey in open habitat with little or no stealth, surveying moving animals for weaknesses, chasing its prey at high speeds and killing it by strangulation.^7,8,9

Several osteological and physiological adaptations of A. jubatus are thought to be related to an extremely fast chase-based predatory behavior.^10,11,12 For example, this species has a highly specialized postcranial skeleton with a number of morphological features that are unique among felids, including the presence of elongated distal limb bones, interphalangeal elastic ligaments, ridged digital pads and lack of fully-retractable claws.^13,14 Indeed, the generic name Acinonyx (from the Greek A-, ‘not’ + kinéō, ‘I move’ +-ónux, ‘claw’), means ‘immobile claws’ referring to the lack of claw hyper-retraction seen in other felids.¹⁵ The lack of fully retractable claws in A. jubatus is thought to be the result of relaxed selective pressures for the protection of the claws from blunting because they are protracted to gain traction during the high-speed chase.^14,16 Moreover, the reduced ability of A. jubatus to manipulate prey is because of its low degree of forepaw dexterity, which ranks among the lowest of felids,^17,18 mainly because of its reduced ability to supinate their forelimbs.^2,19,20,21 As a result, A. jubatus does not manipulate prey with their forepaws.¹⁸ Instead, it causes the prey to lose its balance using its sharp dewclaws on the animal’s skin and shifting its own weight backwards.^14,22,23

These adaptations appear to have evolved to capture small-to-medium sized prey such as Thomson gazelle (Eudorcas thomsonii), which can be subdued with minimal risk of injury and consumed rapidly before surrendering it after the arrival of kleptoparasites.¹² Moreover, A. jubatus possess a reduced skull mass with lightened bone and elongated internal nostrils. This allows it to take a greater volume of air, which is necessary for aerobic exercise during prey chase.^24,25 Not surprisingly, the cheetah shows a greater nasal aperture area than expected from its cranium and palatal dimensions, which evidences a greater breathing capacity compared to other felids,²⁶ and shows also the densest packing of maxilloturbinates among the living cats.²⁷ This is expected in a coursing predator that subdues prey after a prolonged chase, which requires increased breathing capacity for cooling the body during pursuit and compensating the oxygen deficit produced by the huge muscular efforts.²⁸ Moreover, the cheetah’s cranium shows a greater interorbital breadth than a pantherine cat of similar size. A wide braincase is a condition typical of small felids²⁹ and it appears that despite increasing its size to that of a pantherine felid, A. jubatus retained small-cat cranial proportions as it is paedomorphic.²⁸

Compared to the pantherine felids, the cheetah shows some derived features, including (1) a domed and rostro-caudally compressed cranium; (2) a lateral enlargement of the frontals caudally to the zygomatic processes; (3) a widening of the nares and orbits, with the latter oriented frontward; (4) less developed sagittal and nuchal crests; (5) bowed zygomatic arches; (6) slender canines and narrow cheek teeth; and (7) a marked reduction of the protocone in the upper carnassial. In contrast, these features are less marked in the skull of the giant (∼100 kg) cheetah Acinonyx pardinensis from the Eurasian Pleistocene, which suggests that the highly derived skull shape of the modern cheetah probably evolved recently.^24,28,30,31

The few postcranial remains available of A. pardinensis (e.g., a nearly complete foreleg from Dmanisi³²) are suggestive of body proportions similar to the living cheetah (i.e., a slender skeleton with elongated limb segments). This led to the assumption of a direct similarity between the hunting strategy in A. pardinensis and A. jubatus, which would be based on a high-speed chase of small-to-medium sized prey.³² However, the musculoskeletal skull morphology of A. pardinensis suggests that it could catch larger ungulate prey through a killing strategy more similar to the extant pantherine cats than to the cheetah.²⁴ Moreover, a recent study on the inner ear morphology³³ in both A. jubatus and A. pardinensis suggests that the extinct form did not possess the distinctive attributes of the A. jubatus inner ear –i.e., greatest volumes of the vestibular system among cats, a dorsal extension of the anterior and posterior semicircular canals– that presumably correlate with a greater afferent sensitivity of the inner ear to head motions, facilitating postural and visual stability during the high-speed chase.³³

On the other hand, brain overheating is thought to be a limiting factor for chase distance in the cheetah.^6,34 The large frontal sinuses of A. jubatus are highly vascularized, which is thought to provide a cooling mechanism to prevent overheating of the brain during periods of high exertion.^25,35 At the end of a sprint, body temperature reaches ∼38.5–41°C.^6,10,36 After a chase, the cheetah is so exhausted that as many as 30 min may be elapsed catching its breath before it can eat.⁹

On the other hand, A. jubatus also possesses elongated limbs with a reduced muscle mass, which allow it for exerting both faster bursts of speed and longer stride lengths during running.^37,38,39 In addition, a very flexible lumbar spine allows greater extension of the posterior back in the cheetah, which facilitates the aerial and land pose during the gallop phase³⁷ and also increases stride length by 5% and top speeds by 10%.^10,40 The tail of the cheetah can rotate across the horizontal and vertical planes, which allows it to perform conical movements around the sagittal axis of the body to function as a rudder or counterweight. This allows the animal to change its running direction while maintaining balance, and also helps it to prevent skidding and improve aerodynamics.⁴¹

The fossil record attests that this ‘built-for-speed’ design has appeared at least twice within felids: in the Old-World A. jubatus and also in the ‘cheetah-like’ cat Miracinonyx spp. from the North American Pleistocene. As evidenced by both molecular⁴² and morphological⁴³ data, the closest living relative of Miracinonyx is the cougar (Puma concolor). However, its skeletal anatomy is extremely ‘cheetah-like’,^44,45 at least in the more derived M. trumani from the late Rancholabrean.⁴³ The earliest species ascribed to the genus Miracinonyx is M. inexpectatus, which is morphologically different than the more derived M. trumani.⁴³M. inexpectatus differs from M. trumani in a set of traits that evidence a lesser degree of ‘cheetah-like’ specialization compared to the more recent species, including an overall larger size, a longer third upper premolar relative to the fourth upper premolar, a larger protocone in the fourth upper premolar, larger upper canines relative to the upper fourth premolar, smaller nasal aperture area, a stouter distal ulna, a lower brachial index (i.e., ratio of radius length to humerus length), and a less elongated patella.⁴³

It is hypothesized that an ancestral form of Puma-Miracinonyx probably originated in the Old World, migrated to North America about 6 million years ago (Myrs) and gave rise to both Miracinonyx and Puma,^24,43, (but see ref.46) about 4.0 Myrs –a date of divergence that is likely based on the earliest Puma records in North America (P. lacustris) from the Glenns Ferry Formation of Idaho.^47,48 However, this same site also has preserved a single specimen that has been referred to M. inexpectatus, which is likely the oldest record of that species. Therefore, the occurrence of both taxa at a site spanning 4.18–3.11 Ma suggests the early divergence the Miracinonyx and Puma lineages.^47,48

Based on this evidence, the most parsimonious explanation is that the ‘cheetah-like’ morphology of M. trumani is a result of convergent evolution with Acinonyx for fast-pursuit of prey across the prairies and steppe terrains of North America during the Pleistocene.^42,43 Moreover, it has been proposed that Pronghorn ‘antelope’ (Antilocapra americana), the second-fastest modern land mammal with no natural predators that come close to matching their speed, would be the preferred prey of M. trumani.⁴⁹

Further insights on the predatory behavior of M. trumani have been derived from analyses of stable-isotopes abundance of fossil collagen from Natural Trap Cave specimens. The trophic enrichment between herbivores and carnivores in collagen isotopes is 1.6‰ for carbon and 3.9‰ for nitrogen, respectively.⁵⁰ In the case of Natural Trap Cave, the differences in δ¹³C and δ¹⁵N values for the single specimens analyzed by McNulty et al.⁵¹ of M. trumani and A. americana are close to those expected for a carnivore and its prey. However, this also applies to two of the four specimens of bighorn sheep (Ovis canadensis) analyzed, which opens the possibility that M. trumani also preyed on sheep in rocky environments. A more in-depth study of these fossils has been recently performed by Higgins et al.⁴⁹ The mean δ¹³C and δ¹⁵N values obtained for two specimens of pronghorn (−18.7 and 4.7‰, respectively) and five of M. trumani (−17.3 and 7.9‰, respectively) are close –but slightly lower than expected– to the isotopic enrichment from prey to predator. Higgins et al.⁴⁹ confirmed with an isotopic mixing model that pronghorn was the primary prey of M. trumani, contributing to 40% of the diet of the North American cheetah, and that other herbivores such as horse, bison, and sheep were also preyed on.

Although skeletal convergence between M. trumani and A. jubatus has been widely investigated,^{43,45,52,53,54} it is unknown whether the brain architecture of M. trumani is also convergent to that of A. jubatus. The brain of A. jubatus is unique among felids in gyral and sulcal patterns, as well as in regional size and shape, features that have been interpreted as related to its specialized predatory behavior.^55,56,57 Here, we investigate gyral and sulcal patterns of the brain of M. trumani, and also the regional size and shape of functional brain areas from an endocast of a cranium preserved at Natural Trap cave (Northern Wyoming), and compare it quantitatively with a sample of living cats, including its Old-World vicar (A. jubatus) and its closest living relative (P. concolor) (Figure 1). Our main aim is to ascertain whether the brain of M. trumani also possesses the distinctive features present in A. jubatus that are thought to be related to its unique predatory behavior. Our initial hypothesis is that brain architecture of M. trumani is similar to that of A. jubatus given its close skeletal convergence toward a fast-running-chase predatory behavior.

Figure 1.

Brain endocast segmentation

Some crania analyzed are taken as an example. (A) Acinonyx jubatus.

(B) Miracinonyx trumani.

(C) Puma concolor. Brain regionalization is shown in colors following previous studies.⁵⁵

Results

Patterns of gyri and sulci in M. trumani

Following a previous study⁵⁷, the gyral and sulcal patterns of the brain of modern felids is highly conservative, as they do not vary from the largest pantherine felids (Panthera tigris or Panthera leo), with cranial capacities of 250–300 cm³, to the smallest felid species (e.g., Prionailurus rubiginosus), with brains of 20–25 cm³. Moreover, there are not significant differences in regional brain proportions between the largest and smallest species.⁵⁷ However, one notable exception to this pattern is the brain of A. jubatus. For example, its suprasylvian sulcus arch at the caudomedial corner –where the posterior and middle portions of the suprasylvian sulcus join– is disrupted. Our brain endocast data suggest that the suprasylvian sulcus is neither disrupted in P. concolor nor in M. trumani, although the continuation of this sulcus in the latter is more subtle (Figure 2A).

Figure 2.

Patterns of gyri and sulci in P. concolor (left column), M. trumani (middle column) and A. jubatus (right column)

(A) Lateral view of brain endocast showing the pattern of suprasylvian sulcus arch at the caudomedial corner.

(B) Pattern of location of the orbital sulcus and ectosylvian sulcus.

(C) presence or absence of postcruciate sulcus.

Abbreviations: sprs, suprasylvian sulcus; prs, presylvian (orbital) sulcus; an.es, anterior ectosylvian; p.es, posterior ectosylvian; cr.s, cruciate sulcus; pcr.s, postcruciate sulcus. Asterisks denote areas of discussion through the text. See also Figure S1.

Unlike other felids, the orbital sulcus of A. jubatus does not continue with the ectosylvian sulcus, a disruption that is also observed in the Lynx.⁵⁷ The endocast of M. trumani evidences a continuation of the orbital (presylvian) sulcus with the ectosylvian one, although by a very smooth sulcus area (Figure 2B).

The postcruciate sulcus of felids is plainly marked irrespective of brain size,⁵⁷ but in A. jubatus it is usually absent or, at most, is reduced to a small dimple. Our brain endocast data indicates that the postcruciate sulcus is still present in M. trumani as a small dimple, as well as in P. concolor (Figure 2C).

Topological deviations from the brain of M. trumani to the ones of A. jubatus and P. concolor

The brain of A. jubatus has been proposed to be highly globose⁵⁵ and our topological analysis confirms this. Comparing the brain topology of P. concolor with that of M. trumani, the latter have much more developed lateral sides, which may relate to a greater amplification of the motor, somatosensory, auditory, and visual regions of the cortex (Figure 3A). On the other hand, the brain topology of M. trumani is closer (in terms of average distance) to that of P. concolor (0.23 ± 1.76 mm of average distance) than to that of A. jubatus (0.48 ± 1.66 mm of average distance), which probably relates to the more pronounced rostral dorsiflexion of the telencephalon in A. jubatus (Figure 3B). Following the areas defined in the cat brain,⁵⁸ the areas more developed in M. trumani relative to A. jubatus are the prefrontal cortex (PFdm, PFdl), the motor cortex (4γ, 4δ, 6aα, 6aβ), the primary somatosensory cortex (1, 2, 3a, 3b), the fourth somatosensory (S4) and the partially fifth somatosensory cortex (S5), the visual cortex (20a, 21b, 17 and partially 21a, 19) and other areas of the auditory cortex (A2, dPE, iPE, vPE, VAF).

Figure 3.

Topological deviations to the brain of M. trumani from the ones of A. jubatus and P. concolor

(A) Deviations from P. concolor (reference) to M. trumani (target).

(B) Deviations from A. jubatus (reference) to M. trumani (target).

In both cases, warm colors indicate positive deviations (in mm) from the reference to M. trumani, and cold colors indicate negative deviations. See also Figure S2.

Total and regional brain size in M. trumani

The bivariate regression of Total Endocranial volume (TEv) on body mass (BM) was significant (r²= 0.9756; F_(1,13)= 520.18; p<0.0001; Table 1) as well as the regression of both contrasted variables (r²= 0.8828; F_(1,12)= 90.36; p<0.0001). The TEv values for P. concolor as well as those of M. trumani fall closer to the regression line –i.e., within the 95% confidence interval (Figure 4A). On the other hand, our results indicate that the jaguar (Panthera onca) is the less encephalized large cat among the sample (Figure 4A).

Table 1.

Regional brain volumes and body masses of extant felids used in this study

Species	Mass(kg)	TEv(mm3)	ACv(mm3)	PCv(mm3)	CBv(mm3)
A. jubatus	46.7	136,747.818	5687.272	101,852.738	27,644.508
F. silvestris	5.53	39,346.93	2268.24	28,035.93	8268.72
L. geoffroyi	3.59	36,842.94	2026.78	25,939.62	8185.28
L. guigna	2.23	28,505.6	2028.27	19,540.5	6258.3
L. pardalis	11.9	68,282.95	4866.14	46,593.1	14,905.56
L.wiedii	3.25	45,526.52	3371.04	31,506.63	9745.92
L.canadensis	9.37	78,873.34	4999.91	56,902.51	15,459.24
L. rufus	8.91	60,584.94	3145.84	43,061.71	12,817.8
M. trumani	50	152,756.41	7876.76	109,221.88	33,198.75
N. nebulosa	19.676	90,903.89	4861.54	63,650.09	20,728.67
P. leo	161.5	259,987.7567	16,490.56667	183,225.81	53,714.98
P. onca	100	156,479.09	11,540.05	108,563.6	32,399.69
P. pardus	52.038	149,319.62	9954.66	105,203.34	30,425.05
P. tigris	162.56	289,463.72	18,206.34	211,268.89	53,660.62
P. concolor	51.6	147,947.375	7980.15	106,105.9275	31,254.2

Abbreviations: TEv, total endocranial volume; ACv, anterior cerebrum volume; PCv, posterior cerebrum volume; CBv, cerebellum/brain stem volume. Body masses obtained from the literature.^59,60 The body mass of M. trumani was obtained in this study from the regression of mass against skull length for modern felids.⁶¹ The data represent species averages computed from this study and from previous data. See method details and quantification and statistical analysis.⁵⁵

Figure 4.

Total and regional brain size in M. trumani

(A) Total endocranial volume (in mm³) against body mass (in kg), both variables log-transformed.

(B) Anterior cerebrum volume (in mm³) against total endocranial volume (in mm³), both variables log-transformed.

Abbreviations: Lguig, Leopardus guigna; Lwie, Leopardus wideii; Fgeo, Felis geoffroyi; Fsil, Felis silvestris; Lruf, Lynx rufus; Lpar, Leopardus pardalis; Lcan, Lynx canadensis; Nneb, Neofelis nebulosa; Ajub, Acinonyx jubatus; Ppar, Panthera pardus; Ponc, Panthera onca; Pcon, Puma concolor; Mira, Miracinonyx trumani; Ptig, Panthera tigris; Pleo, Panthera leo.

The bivariate regression of the Anterior Cerebrum volume (ACv) against TEv shows a significant association (r²= 0.9485; F_(1,13)= 239.5; p<0.0001; Table 1), even when taking into account the phylogenetic relationships of the species (r²= 0.9187; F_(1,12)= 135.7; p<0.0001). Strikingly, although the ACv of P. concolor is not reduced relative to its TEv compared to other felids –it falls within the 95% confidence interval of the regression line– the ACv of M. trumani is slightly reduced to its estimated TEv, falling outside the 95% confidence interval below the regression line (Figure 4B). Despite this, the value of ACv for M. trumani is closer to the one of P. concolor than to the one of A. jubatus (Figure 4B).

Paranasal sinuses size in M. trumani

The relative paranasal sinuses volume to total cranium volume is shown in Figure 5. Our results indicate that all specimens of A. jubatus possess a paranasal sinuses volume surpassing the 25% of cranium volume. In contrast, the paranasal sinuses volume of M. trumani and P. concolor varies between 15 and 20% of total cranium volume (Figure 5).

Figure 5.

Paranasal sinuses size in M. trumani

Histogram showing the ratio between paranasal sinuses volume and cranium volume. 1-5: A. jubatus; 6-8: P. concolor; 9,10: P. leo; 11: L. rufus; 12: N. nebulosa; 13: M. trumani. Abbreviations: Ps v, paranasal sinuses volume; Skl v, skull (cranium) volume. Paranasal 3D models are not scaled. Silhouettes obtained from Phylopic (phylopic.org).

General brain shape in M. trumani

The bivariate plot depicted by the first two eigenvectors obtained from a PCA of the 20 landmarks digitized (Figure 6) on the brain endocasts is shown in Figure 7. Inspection of other PCs do not reveal a clear pattern of specimen ordination, as they highlight few taxa. The first PC, which explains 34.35% of the original variance, mainly separates A. jubatus from the other felids. The morphological variation accounted for by this eigenvector is mainly related to the antero-posterior location of the most-lateral point of the frontal lobe (i.e, landmarks 4,5) and the mid-sagittal brain (i.e, landmarks 2) (Figure 7).

Figure 6.

Landmarks digitized from brain endocasts for geometric morphometric analyses

(A) Left lateral view. (B) posterior view. (C) right lateral view. (D) frontal view. Landmarks are defined as follows: 1: mid-sagittal point in frontal location; 2: mid-sagittal point in dorsal location; 3: mid-sagittal in caudal location; 4: most-lateral point of right frontal lobe; 5: most-lateral point of left frontal lobe; 6: most-lateral point of right temporal lobe; 7: lowest point of right temporal lobe; 8: most-lateral point of left temporal lobe; 9: lowest point of the left temporal lobe; 10: point of maximum curvature of the right suprasylvian gyrus; 11: point of maximum curvature of the left suprasylvian gyrus; 12: point of maximum curvature of the right ectosylvian gyrus; 13: point of maximum curvature of left ectosylvian gyrus; 14: most-posterior point of the right occipital gyrus; 15: most-posterior point of the left occipital gyrus; 16: deepest point of the right sylvian sulcus; 17: deepest point of the left sylvian sulcus; 18; most-lateral point of the right cruciate sulcus; 19: most-lateral point of the left cruciate sulcus; 20: midpoint of the cruciate sulcus.

Figure 7.

General brain shape in M. trumani

Morphospace depicted by the first two PCs obtained from a Principal Component Analysis of 20 landmarks digitized in 13 brain endocasts. Silhouettes obtained from Phylopic (phylopic.org). See also Figure 6.

The second PC, which explains 20.36% of the original variance, mainly separates among the felids P. leo and M. trumani, which both take positive scores, from P. concolor, which score with intermediate values, and Neofelis nebulosa and Lynx rufus, which both take negative projections (Figure 7). The morphological variation accounted for by this eigenvector relates to the antero-posterior location of the cruciate sulcus (i.e., landmarks 18,19), the most-lateral (i.e., landmarks 6,7) and lowest (i.e., landmarks 8,9) points of the temporal lobe, as well as the maximum curvature of the right and left suprasylvian gyrus (i.e., landmarks 10,11) (Figure 7).

Discussion

Our results indicate that, compared to A. jubatus, the suprasylvian sulcus is not disrupted in either M. trumani or P. concolor, although the continuation of this sulcus in M. trumani is subtle. Strikingly, this disruption has been interpreted as a consequence of the expansion of the Clare-Bishop area, which probably relates to a visual specialization in A. jubatus, a diurnal predator that relies on eyesight for prey detection and chase.⁵⁷ Our results for M. trumani may tentatively indicate that it also relied on eyesight for detecting prey, but to a lesser degree than the cheetah. Moreover, the orbital sulcus of M. trumani does slightly continue with the ectosylvian sulcus, a pattern that is different from the one observed in A. jubatus. The interruption of orbital-ectosylvian sulci in the cheetah relates to its more globose brain, which in turn relates to the position of the anterior coronal and anterior suprasylvian gyri that bulge out more beyond the lateral boundary of the sigmoid gyri than in other similar-sized felids.⁵⁷ Therefore, the continuation of the orbital-ectosylvian sulci in M. trumani reflects that its brain is more globose than the one of P. concolor, although it does not reach the extreme degree observed in A. jubatus. This is also confirmed by our topological analysis of the brain of M. trumani. However, our results indicate that compared to the brain of P. concolor, the brain of M. trumani is characterized by the presence of a well-developed somatosensory cortex and visual areas, as well as by the expansion of other areas related to the auditory cortex. Although this may indicate that M. trumani has enhanced visual and auditory functions compared to P. concolor, it also exhibits a greater motor complexity than A. jubatus, the latter probably related to the presence of fully retractable claws in M. trumani.⁴³ In any case, future quantitative studies of brain functional areas will clarify which brain functions in M. trumani are enhanced or diminished compared to both Puma and Acinonyx.

Following a previous study,⁵⁷ the postcruciate sulcus separates the primary motor and somatosensory cortical areas that control the postcranial part of the body. The lack of that sulcus in A. jubatus has been interpreted as reflecting its relatively small motor cortex, which would be related to the presence of a less developed limb musculature.^55,57 The degree of forepaw dexterity in A. jubatus ranks the lowest among the modern felids,¹⁸ mainly because of its reduced ability to supinate the forelimb.^{2,19,20,21,62} As a result, its ability to climb trees and manipulate prey is diminished compared to other taxa.⁶³ This likely relates to its specialized predatory behavior, which is based in a fast-running chase at the expense of the loss of manipulatory capabilities with the forelimb.^14,64 Unlike other felids, A. jubatus lacks fully retractable claws, which improves limb traction and support.¹⁴ Most cats use their claw-equipped forelimbs to grapple and manipulate prey. This is especially relevant for large-sized species, which typically take prey with a body size that equals to, or is greater than, their own.⁶⁵A. jubatus is the only large cat that regularly kills prey with a body weight that is less than its own^16,66,67 and its killing bite is performed at the ventral region of the neck by strangulation, bringing down prey by hooking one of its dewclaws into the animal and shifting its own weight posteriorly.^14,22,68,69 Indeed, it has been proposed that although the dewclaw in the tiger (P. tigris), lion (P. leo) and leopard (Panthera pardus) is only slightly larger than the claw of the second digit, the dewclaw of the cheetah is enlarged relative to other digits, a condition that should be related to its separate role.²³ On the other hand, the dewclaw of P. concolor was intermediate in size to that of pantherines and that of A. jubatus, and Londei²³ explained the condition of P. concolor as a leftover trait that was probably inherited from more cursorial (extinct) forms. However, although the size of the dewclaw could be a key trait for the interpretation of the predatory behavior of M. trumani, whether digit I is also enlarged in this species remains to be investigated.

It is worth noting that the postcruciate sulcus is also present in the canids, which do not possess retractable claws (excepting Urocyon cinereoargenteus) and use their forelimbs almost exclusively for running.^2,20,64 However, the postcruciate sulcus appeared independently in canids and felids.⁷⁰

It is surprising that even though M. trumani is extremely cheetah-like in appearance and limb proportions, it shows fully-retractable claws. This could explain the presence in this predator of an expanded motor cortex compared to A. jubatus. In turn, this could relate to its greater ability for prey manipulation and support, which could explain the retention of M. trumani (as in P. concolor) of the postcruciate sulcus as a small dimple. However, future ecomorphological studies of M. trumani based on its major limb bones could give some clues on this topic, as ecomorphological studies performed on the appendicular skeleton has provided proxies of predatory behavior in the living carnivorans.^{2,13,17,20,21,71,72,73,74,75,76,77}

On the other hand, the brain of A. jubatus is the smallest relative to its body mass among the living felids,⁵⁵ which has been interpreted as an adaptation for weight loss and energy saving, aspects that might be advantageous for a predatory behavior based on a fast-running chase. If this explanation holds true, the endocranial volume for M. trumani could tentatively indicate that this predator was not as equipped for fast-running as A. jubatus. However, our data suggests that the TEv of A. jubatus is not more reduced than that of other felids such as the jaguar (P. onca), which is a generalized predator.⁵⁹ However, it is worth noting that the body mass of P. onca is proportionately elevated, rather than having reduced TEv relative to body mass.

Strikingly, the volume of the anterior cerebrum of A. jubatus, which apparently relates to the degree of sociality in carnivores, is highly reduced compared to other felids⁵⁵–even though males can form coalitions of 2 or 3 related individuals,^78,79 they do not cooperate during a hunt.⁸⁰ However, it is worth noting that the relatively small brain of A. jubatus could be also explained by its low genetic diversity because of past population bottlenecks,^81,82,83 although we find this possibility speculative. In any case, the anterior cerebrum of M. trumani is also reduced compared to P. concolor, but without reaching the extreme reduction seen in A. jubatus.

Apart from these differences, the brain of A. jubatus also differs from the other felids in general shape. For example, it has been noted that the brain of the cheetah shows a unique rostral dorsiflexion among felids.⁵⁵ Our geometric morphometric analysis, based on 3D landmarks digitized from relative positions of gyri and sulci, indicates that the brain of A. jubatus possesses more posteriorly positioned frontal lobes and more anteriorly positioned mid-sagittal brains than in other felids. This overally relates to the brain dorsiflexion of the cheetah.⁵⁵ In this respect, the brain of M. trumani does not significantly differ from that of other felids, and therefore, it is not characterized by having the dorsiflexion typical of A. jubatus.

Although, fitting the brain into the bony skull with its functions requires integration between the two parts^84,85,86 and the ‘spatial packing hypothesis’ points that spatial constraints might shape brains to be packed more globularly when their mass increases relative to body mass,⁸⁷ the overall cranium shape of Acinonyx and M. trumani is very similar and rather distinct of other felid species.⁵² Therefore, the influence of external cranium shape on endocast form could be in principle negligible. This could be extended to cranium function (e.g., feeding) because the latter is reflected on cranium shape in carnivores.^{88,89,90,91,92} However, this rostral dorsiflexion of the brain of A. jubatus reflects the presence of enlarged frontal sinuses,²⁵ which leads to its highly domed cranium.²⁸ The enlargement of the frontal sinuses is thought to act as a vascular cooling mechanism during high-speed chases.^25,35 Indeed, at the end of a sprint, the body temperature of A. jubatus is ∼41°C^6,10 and the large frontal sinus plays an important role in preventing brain overheating.²⁵ Again, our qualitative assessment of paranasal sinuses volume in M. trumani indicates that it did not possess the enlarged sinuses typical of A. jubatus. Therefore, brain overheating was probably not a selective agent here. This could be related to the fact that M. trumani did not perform the bursts of speed of a cheetah, or could even reflect that Pleistocene temperatures in North America were much lower than today in Africa and Iran.^93,94 In any case, the absence of well-developed sinuses in M. trumani is probably behind the lack in this felid of the brain rostral dorsiflexion typical of A. jubatus.

In summary, all these morphological traits seem to indicate that M. trumani was not as specialized as A. jubatus in deploying a fast-running pursuit or, at least, that it was not ‘cognitively’ equipped for this predatory behavior. Therefore, given the skeletal resemblance between M. trumani and A. jubatus, it is reasonable to think that the skeletal adaptations for fast-running evolved faster or earlier than those related with the architecture of the brain. However, our results also indicate that M. trumani does not possess paranasal sinuses as developed as the living cheetah –although this trait has been traditionally used to justify the convergent evolution between M. trumani and A. jubatus.^44,45 This evidence, together with the presence of fully-retractable claws in M. trumani, cast doubts that M. trumani deployed a fast-running chase as specialized as in A. jubatus. However, although the absence of fully-retractable claws has been interpreted as an adaptation to enhance limb traction and support during a chase,¹⁴ this interpretation could be biased for the absence of retractable claws in the pack-hunting canids. To us, it is not really clear whether A. jubatus lost its fully retractable claws as a response of its highly specialized predatory behavior based in fast-pursuit of open-country prey, or it was just a consequence of preying upon small prey. If the second possibility holds, the presence of retractable claws in M. trumani would not be an argument to support the hypothesis that it was not as specialized as A. jubatus for fast running. Instead, it would indicate that it preyed upon larger prey.

On the other hand, it has been found remains of M. trumani in different caves from the Grand Canyon of northern Arizona and proposed that M. trumani should be envisaged as a species better adapted to dry uplands and rocky canyons, and not restricted to savanna-like settings.⁹⁵ In this respect, it is claimed that the ecology of the Grand Canyon M. trumani was similar to the living Asiatic cheetah (Acinonyx jubatus venaticus) and snow leopard (Panthera uncia), which are adapted for pursuit of mountain and canyon ungulates over near vertical rocky and mountainous terrain.⁹⁵

In any case, our results were totally unexpected given the overall skeletal (and dental) resemblance between A. jubatus and M. trumani, but ecomorphological studies from limb bones in M. trumani could answer whether the skeleton of this predator was already equipped for a fast-running chase and whether skeletal modification preceded the evolution of brain architecture in this formidable lineage of big cats.

Limitations of the study

Our study is based on a single endocast virtually extracted from a cranium preserved in Natural Trap Cave. As a result, we lack the broader perspective offered by large intraspecific studies. Moreover, following previous studies⁵⁷ we assumed that the distinctive traits of A. jubatus brain are related to its predatory behavior, but this could also relate to its genetic impoverishment because of the severe population bottlenecks experienced by this species in the past.

Another caveat we find in our study is the possibility that the brain of A. jubatus could be (in part) the result of its expanded sinuses. Future studies investigating the role of sinus development in A. jubatus to explain the characteristic dorsiflexion of its brain are necessary to confirm or refute whether the brain architecture of A. jubatus is partially a by-product of other internal structures such as the sinuses.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Biological samples
Miracinonyx trumani	University of Kansas Natural History Museum	KUVP-51277
Acinonyx jubatus	University of Wisconsin Zoological Collection	UWZS 23961
Acinonyx jubatus	Field Museum of Natural History (Chicago, USA)	FMNH 29635
Acinonyx jubatus	Field Museum of Natural History (Chicago, USA)	FMNH 127834
Acinonyx jubatus	Osteological Museum of Valladolid University	VU 6075
Acinonyx jubatus	Osteological Museum of Valladolid University	VU 6394
Puma concolor	University of Wisconsin Zoological Collection	UWZS 32281
Puma concolor	Osteological Museum of Valladolid University	VU 409
Puma concolor	Osteological Museum of Valladolid University	VU 3087
Panthera leo	Osteological Museum of Valladolid University	VU 6080
Panthera leo	Osteological Museum of Valladolid University	VU 2685
Lynx rufus	Ohio University Vertebrate Collection	OUVC 9576
Neofelis nebulosa	National Museum of Natural History, Washington DC (USA)	USNM 282124
Deposited data
Endocast models	Figshare	Data from: Built for speed? The brain of the North American cheetah-like cat Miracinonyx trumani, https://figshare.com/s/b1366f7e7aef6317c62e
Scanning parameters	supplemental information	Table S1
Software and algorithms
MorphoJ	Klingenberg96	https://morphometrics.uk/MorphoJ_page.html
Rstudio	R core team97	https://www.rstudio.com/
3D-slicer	Kikinis et al.98	https://www.slicer.org/
Fiji (ImageJ)	Schindelin99	https://imagej.net/software/fiji/
Geomagic studio	3D Systems100	https://es.3dsystems.com/software
Geiger package	Harmon et al.101	https://cran.r-project.org/web/packages/geiger/index.html
Mesquite	Maddison and Maddison102	https://www.mesquiteproject.org/
Ape	Paradis et al.103	https://cran.r-project.org/web/packages/ape/index.html

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Dr. Borja Figueirido (Borja.figueirido@uma.es).

Materials availability

• •This study did not generate unique reagents.

Experimental model and subject details

CT-scanning of data

We CT scanned a complete cranium of M. trumani (KUVP-51277) unearthed from Natural Trap Cave (northern Wyoming, USA) with an age of ca. 23-25 ka⁹⁶. Additional skulls of other felid species were also CT-scanned: 5 cheetahs (A. jubatus), 2 lions (Panthera leo), 3 cougars (Puma concolor), 1 clouded leopard (Neofelis nebulosa), and 1 bobcat (Lynx rufus). See Table S1.

Cranium KUVP-51277 of Miracinonyx trumani was unearthed from Natural Trap Cave (northern Wyoming, USA) with an age of ca. 23-25 ka, Late Pleistocene¹⁰⁴ and it is housed in the University of Kansas Vertebrate Paleontology (KUVP). The cranium of this specimen was CT-scanned at University at the Wisconsin Institute for Medical Research’s Imaging Services Department of the University of Wisconsin with a GE Medical System Discovery model CT750 in Helicoidal mode. The acquisition parameters were: Kvp 120; X-ray tube current 250A (Ampere); slice thickness 0.625; spacing between slices 0.3120; exposure time 912; field of reconstruction 143 mm; pixel spacing 0.2793 mm; pixel size 512x512 and voxel size 0.2793 X 0.2793 X 0.3120 mm. Number of images in set 2974 in 16-bit TIFF images. See Table S1.

Cranium UWZS 23961 of Acinonyx jubatus belongs to a captive female from William Lowe Game Farm, Beaver Dam, Dodge County, Wisconsin, USA. This specimen was scanned at the Wisconsin Institute for Medical Research’s Imaging Services Department (University of Wisconsin). The CT scan machine used is a GE Medical System Discovery CT750. The CT scanner was done in Helicoidal mode. The acquisition conditions were as follows: Kvp 120; X-ray tube current 250 A (Ampere); slice thickness 0.625; spacing between slices 0.3122; exposure time 912; field of reconstruction 222 mm; pixel spacing 0.4336 mm; pixel size 512x512 and voxel size 0.4336 X 0.4336 X 0.3122 mm. Number of images in set 1848 in 16-bit TIFF images. See Table S1.

Cranium FMNH 29635 of Acinonyx jubatus raineyii is a male of wild origin from Eastern Kenya housed in the Field Museum of Natural History of Chicago (USA) and it was CT-scanned at the University of Texas High-Resolution X-ray CT-Facility using the following parameters: Kvp 420; X-ray tube current 180 A (Ampere); slice thickness 0.5 mm; inter slices 0.48 mm; field of reconstruction 131 mm; pixel spacing 0.26 mm; pixel size 512x512 and voxel size 0.26 X 0.26 X 0.48 mm. Number of images in set 393 in 16-bit TIFF images. See Table S1.

Cranium FMNH 127834 of Acinonyx jubatus raineyii belongs to a female of wild origin from Kenya and it is housed in the Field Museum of Natural History (Chicago, USA). This specimen was scanned at the University of Texas High-Resolution X-ray CT Facility. The acquisition conditions were: Kvp 420; X-ray tube current 180 A (Ampere); slice thickness 0.5 mm; inter slices 0.46 mm; field of reconstruction 131 mm; pixel spacing 0.25 mm; pixel size 507x402 and voxel size 0,25 X 0.25 X 0.46 mm. Number of images in set 363 in 16-bit TIFF images. See Table S1.

Cranium VU 6075 of Acinonyx jubatus belongs to a captive male and it is housed in the Osteological Museum of Valladolid University. Such specimen was scanned at Vithas Medical Center (Malaga, Spain). The skull was CT-scanned with a GE Medical Systems (Brivo CT385 Series) and the conditions of acquisition were: Kvp 120; X-ray tube current 95 A (Ampere); slice thickness 0.625mm; inter slices 0.31 mm; field of reconstruction 265 mm; pixel spacing 0.517 mm; pixel size 512x512 and voxel size 0.517 X 0.517 X 0.31 mm. Number of images in set 1234 in 16-bit TIFF images. See Table S1.

Cranium VU 6394 of Acinonyx jubatus belongs to a captive female and it is housed in the Osteological Museum of Valladolid University. Such specimen was scanned at Vithas Medical Center (Malaga, Spain). The skull was CT-scanned with a GE Medical Systems (Brivo CT385 Series) and the conditions of acquisition were: Kvp 120; X-ray tube current 95 A (Ampere); slice thickness 0.625mm; inter slices 0.31 mm; field of reconstruction 265 mm; pixel spacing 0.517 mm; pixel size 512x512 and voxel size 0.517 X 0.517 X 0.31 mm. Number of images in set 1234 in 16-bit TIFF images. See Table S1.

Cranium UWZS 32281 of Puma concolor belongs to a wild animal from Wyoming, USA. This specimen was scanned at the Wisconsin Institute for Medical Research’s Imaging Services Department (University of Wisconsin). The CT scan machine used is a GE Medical System Discovery CT750. The CT scanner was done in Helicoidal mode. The acquisition conditions were: Kvp 120, X-ray tube current 250 A (Ampere), slice thickness 0.625, spacing between slices 0.3122, exposure time 912, field of reconstruction 179 mm; pixel spacing 0.3496 mm, pixel size 512x512 and voxel size 0.3496 X 0.3496 X 0.3122 mm. Number of images in set 2862 in 16-bit TIFF images. See Table S1.

Cranium VU 409 of Puma concolor belongs to a captive male and it is housed at the Osteological Museum of Valladolid University. Such specimen was scanned at Vithas Medical Center (Malaga, Spain). The skull was CT-scanned with a GE Medical Systems (Brivo CT385 Series) and the conditions of acquisition were: Kvp 120; X-ray tube current 114 A (Ampere); slice thickness 0.625mm; inter slices 0.31 mm; field of reconstruction 250 mm; pixel spacing 0.4883 mm; pixel size 512x512 and voxel size 0.4883 X 0.4883 X 0.31 mm. Number of images in set 1040 in 16-bit TIFF images. See Table S1.

Cranium VU 3087 of Puma concolor belongs to a captive female and it is housed in the Osteological Museum of Valladolid University. Such specimen was scanned at Vithas Medical Center (Malaga, Spain). The skull was CT-scanned with a GE Medical Systems (Brivo CT385 Series) and the conditions of acquisition were: Kvp 120; X-ray tube current 80 A (Ampere); slice thickness 0.625mm; inter slices 0.31 mm; field of reconstruction 367 mm; pixel spacing 0.7168 mm; pixel size 512x512 and voxel size 0.7168 X 0.7168 X 0.31 mm. Number of images in set 1195 in 16-bit TIFF images. See Table S1.

Cranium VU 6080 of Panthera leo belongs to a captive male and it is housed in the Osteological Museum of Valladolid University. Such specimen was scanned at Vithas Medical Center (Malaga, Spain). The skull was CT-scanned with a GE Medical Systems (Brivo CT385 Series) and the conditions of acquisition were: Kvp 120; X-ray tube current 106 A (Ampere); slice thickness 0.625mm; inter slices 0.31 mm; field of reconstruction 319 mm; pixel spacing 0.6231 mm; pixel size 512x512 and voxel size 0.6231 X 0.6231 X 0.31 mm. Number of images in set 1161 in 16-bit TIFF images. See Table S1.

Cranium VU 2685 of Panthera leo belongs to a captive female housed in the Osteological Museum of Valladolid University. Such specimen was scanned at Vithas Medical Center (Malaga, Spain). The skull was CT-scanned with a GE Medical Systems (Brivo CT385 Series) and the conditions of acquisition were: Kvp 120; X-ray tube current 75 A (Ampere); slice thickness 0.625mm; inter slices 0.31 mm; field of reconstruction 288 mm; pixel spacing 0.5625 mm; pixel size 512x512 and voxel size 0.5625 X 0,5625 X 0.31 mm. Number of images in set 1450 in 16-bit TIFF images. See Table S1.

Cranium OUVC 9576 of Lynx rufus belongs to a wild specimen and it is housed in the Ohio University Vertebrate Collection. The CT-scanner was a General Electric eXplore Locus in vivo, Ohio University MicroCT Facility (Ohio University). The acquisition conditions were: Kvp 80; X-ray tube current 450 A (Ampere); inter slices 0.09 mm; pixel spacing 0.09 mm; pixel size 1072x728 and voxel size 0.09 X 0.09 X 0.09 mm. Number of images in set 1580 in 16-bit TIFF images. See Table S1.

Cranium USNM 282124 of Neofelis nebulosa belongs to a captive male and it is housed in Smithsonian Institution, National Museum of Natural History, Washington DC (USA). This specimen was scanned at the University of Texas High-Resolution X-ray CT Facility. The acquisition conditions were: Kvp 419; X-ray tube current 180 A (Ampere); slice thickness 0.25 mm; inter-slices 0.25 mm; field of reconstruction 194 mm; pixel spacing 0.1890 mm; pixel size 1024x1024 and voxel size 0.1890 X 0.1890 X 0.25 mm. Number of images in set 353 in 16-bit TIFF images. See Table S1.

Method details

Segmentation of brain endocasts

Data were saved as Digital Imaging and Communications in Medicine Centricity (DICOM) in 16 bits and imported into 3D-slicer⁹⁸ to obtain 3D endocasts. An improvement of the definition of the internal structures and the topology of the brain endocast was performed using the resampling tool in ImageJ using Fiji⁹⁹ following previous studies.¹⁰⁵

Following previous studies,⁵⁵ we selected the empty endocranial space in each coronal section from the cribriform plate to the opening of the foramen magnum. Afterwards, we compiled all slices to render a 3D virtual endocast. We also segmented the paranasal sinuses in our 12 specimens to calculate paranasal sinuses volume, in parallel with the volume of the brain endocast.

The extracted endocasts are available at Figshare (https://figshare.com/s/b1366f7e7aef6317c62e).

Investigation of gyri and sulci

To investigate the pattern of gyri and sulci in M. trumani from its brain endocast, we followed the terminology of previous authors^57,58,106 as brain maps to identify those gyri and sulci in our sample, especially in A. jubatus, M. trumani and P. concolor. Moreover, as the gyri and sulci are not easily identifiable in those brain endocasts obtained from medical CTs because they can be blurred, we applied the curvature map in Geomagic Essentials,¹⁰⁰ which quantifies the mean curvature (MC) as the average value between maximal and minimal curvatures in local surfaces (see Figure S1). The value of MC allows to classify the vertices and generate a color pattern based on concave (MC<0, blue), convex (MC>0, red) and flat (MC=0, green) surfaces.^105,97

Quantification and statistical analysis

We performed two topological analyses to compare quantitatively the endocast of M. trumani (target) with that of P. concolor and A. jubatus. To eliminate size effects in both comparisons, the target model was scaled to that of reference models for each comparison. The target endocast was increased by a scale factor of 5% (M. trumani vs. P. concolor), and reduced in size by a factor of 5% (M. trumani vs. A. jubatus) (Figure 4). We also quantified topological deviation between the target and reference models in each comparison using an alignment with the minimum average distance models with the software Geomagic essentials (https://es.3dsystems.com/software).See also Figures S1 and S2.

To do this, we fitted the two meshes for each comparison so that the distance among them should be the minimum. Accordingly, the target model is scaled to the reference model. This is necessary to perform a better pairwise superimpositions using the best fit algorithm during the alignment.

The alignment algorithm matches the coordinates of the target model to the reference model. The deviation between both geometric shapes is minimized within a tolerance of <1mm. This process uses the options of exhaustive symmetry and high-precision assembly algorithm, through a point-to-point resampling between the two topologies, choosing the closest points in each iterative process. Each inspection was carried out under a sampling relation of 100% and a maximum iteration count of 5000. The average positive and negative deviations, standard deviations, and the root-mean-square mean square values were obtained from each deviation output of Euclidean distance between the target and reference models.

From these data, we quantified the topological deviation between the target and reference models in each comparison using an alignment with the minimum average distance models previously explained (Figures S1 and S2). In the first topological analyses, we compared M. trumani with P. concolor (Figures S1 and S2), obtaining an average distance of 0.23 mm between the two models, a standard deviation of 1.75 mm and a root mean square of 1.177. In the second topological analyses, we compared the brain endocast of M. trumani with that of A. jubatus (Figures S1 and S2), obtaining an average distance of 0.48 mm, a standard deviation of 1.66 and a root mean square of 1.73 mm.

Quantifying regional and total brain sizes

We used osteological landmarks and gyral/sulcal patterns following previous authors⁵⁵ to subdivide the endocast into four brain regions (olfactory bulb [OB], anterior cerebrum [AC], posterior cerebrum [PC], and cerebellum/brain stem [CB+BS], –i.e. cerebellum plus medulla, pons and part of the caudal mesencephalon) (Figure 1) to calculate regional volumes of these brain regions (OBv, ACv, PCv, CB+BSv). Total endocranial and regional volumes –as well as paranasal sinuses volume– were obtained using Geomagic Essentials¹⁰⁰ and are shown in Table 1. Moreover, the calculated volumes of these regions were summed to obtain a total endocranial volume (TEv).

We regressed the total endocranium volume (TEv) against the body mass (BM) of each species (Table 1) to investigate whether M. trumani possessed an endocranial volume comparable to the one of A. jubatus. The BMs of all living felid species were taken from the literature.⁶⁰ The body mass of M. trumani was calculated from the regression equation of skull length against body mass of modern felids.⁶¹ Skull length of M. trumani skull was measured with Geomagic Essentials¹⁰⁰ from the 3D model and gave 17.8 cms. Moreover, given that other authors⁵⁵ observed that the brain of A. jubatus is characterized by having a significantly lower ACv relative to TEv, we also regressed these variables. In both cases, we used Ordinary LeastSquares regression analysis of log-transformed data with R studio.¹⁰⁷ We also tested whether these associations were biased by phylogenetic inheritance (inflation of Type I error) with Independent Contrast Analysis¹⁰¹ using the package Geiger¹⁰⁸ of the R studio.¹⁰⁷ To do this, we used a carnivoran supertree¹⁰² and we added Miracinonyx from previous sources⁴² and using Mesquite.¹⁰³ Those species not present in our dataset were pruned using the R package Ape.¹⁰⁹ In order to maximize sample size in bivariate regression analyses and not compromise statistical significance due to low sample sizes, we used the values of TEv and ACv for the species sampled by previous authors⁵⁵ and we averaged the values of TEv and ACv for those species present in both datasets.

Our limited CT dataset precluded us to compute a bivariate regression analysis between paranasal sinuses volume against skull volume in order to explore quantitively whether M. trumani exhibits an enlarged paranasal sinus as in the case of A. jubatus. Therefore, we calculated a ratio between both variables and we qualitatively explored sinus volume in M. trumani relative to other felids present in the sample.

Quantifying brain shape

To quantify brain shape differences between M. trumani and other felid species, we digitized a set of 20 landmarks in 3D from 13 felid endocast (Figure 6). The landmarks were subject to Procrustes superimposition¹¹⁰ to remove the effects of size, rotation, and translation. The Procrustes coordinates were subject to Principal Components Analysis (PCA). To evaluate allometric effects, we regressed the Procrustes coordinates onto the logarithm of Centroid size using multivariate regression analysis.⁹⁶ Afterwards, Procrustes coordinates and size-free residuals of the aforementioned regression were both analyzed with Principal Components Analyses using covariation. The geometric morphometric analyses were computed using MorphoJ.¹¹¹

Published: December 22, 2022

Data and code availability

• •The 3d endocast generated in this study have been deposited to Figshare: https://figshare.com/s/b1366f7e7aef6317c62e.
• •Data of regional and total brain volumes and species body masses are provided in Table 1.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.
• •This paper does not report original code.