Understanding the pattern of brain evolution in early rodents is central to reconstructing the ancestral condition for Glires, and for other members of Euarchontoglires including Primates. We describe the oldest virtual endocasts known for fossil rodents, which pertain to Paramys copei (Early Eocene) and Paramys delicatus (Middle Eocene). Both specimens of Paramys have larger olfactory bulbs and smaller paraflocculi relative to total endocranial volume than later occurring rodents, which may be primitive traits for Rodentia. The encephalization quotients (EQs) of Pa. copei and Pa. delicatus are higher than that of later occurring (Oligocene) Ischyromys typus, which contradicts the hypothesis that EQ increases through time in all mammalian orders. However, both species of Paramys have a lower relative neocortical surface area than later rodents, suggesting neocorticalization occurred through time in this Order, although to a lesser degree than in Primates. Paramys has a higher EQ but a lower neocortical ratio than any stem primate. This result contrasts with the idea that primates were always exceptional in their degree of overall encephalization and shows that relative brain size and neocortical surface area do not necessarily covary through time. As such, these data contradict assumptions made about the pattern of brain evolution in Euarchontoglires.
Rodentia is the most taxonomically diverse mammalian order, including 2277 living species divided into 33 families . Extant rodents are also extremely variable in brain shape and size . Since the brain does not preserve in the fossil record, endocasts (representations of the inside of the cranial cavity) are essential to understanding how this diversity emerged. We describe virtual endocasts obtained via high-resolution X-ray computed tomography (CT) for Paramys copei  (Early Eocene [Wasatchian], Wind River Formation, Wyoming) and Paramys delicatus  (Middle Eocene [Bridgerian], Green River Formation, Wyoming). Paramys is one of the first genera of rodents to occur in North America and is known from specimens that are among the oldest for fossil rodents worldwide . These exceptionally preserved endocasts shed light on the earliest stages of rodent brain evolution and permit detailed comparisons with later occurring rodents , as well as other Palaeocene and Eocene euarchontoglirans [7–12].
2. Description and comparisons
Both specimens have a well-preserved braincase with no obvious deformations but unlike Pa. delicatus (AMNH 12506), Pa. copei (AMNH 4756) lacks the zygomatic bones and the rostrum (figure 1). Detailed descriptions of both crania have previously been published [3,4,13,14]. The specimen AMNH 4756 was found in the Lost Cabin Member, Wind River Formation, WY (Wasatchian North American Land Mammal Age [NALMA]; Wa7; 52.4–50.1 Ma)  while AMNH 12506 was discovered in the Blacks Fork Member, Green River Formation, WY (Bridgerian NALMA, Br2, 49.2–47 Ma) . The method for reconstruction of the endocasts can be found in the electronic supplementary material, SI Text. Comparisons were made to the only previously published virtual endocasts for rodents, which pertain to Early Oligocene (Orellan NALMA) Ischyromys typus and extant Sciurus carolinensis , as well as stem primates (plesiadapiforms) and members of the extinct euarchontogliran family Apatemyidae [7–12].
The volume of the olfactory bulbs corresponds to 6.05% of the total volume of the endocast for Pa. copei and 4.74% for Pa. delicatus (table 1). Both species have larger olfactory bulbs compared to Is. typus (e.g. ROMV 1007), S. carolinensis (AMNH 258346)  and plesiadapiform primates such as Microsyops annectens [8–10] (electronic supplementary material, table S1); however, Pa. delicatus (AMNH 12506) and Pa. copei (AMNH 4756) have smaller olfactory bulbs compared with the apatemyid Labidolemur kayi  (table 1; electronic supplementary material, tables S1 and S2). Both species of Paramys have smaller olfactory bulbs than would be expected for their brain volume compared with archaic mammals, but larger than would be expected for fossil and extant Euprimates and Sciuromorpha (electronic supplementary material, figure S1a and table S5). This may suggest a decrease in olfactory bulb relative size occurring through time in Rodentia, independently from Primates. Unfortunately, no quantitative data have been reported for the primitive member of Glires, Rhombomylus turpanensis , so more data on early representatives of Glires are necessary to determine when this reduction might have begun. Interestingly, both species of Paramys have smaller olfactory bulbs than would be expected based on their body mass compared with archaic mammals and extant sciuromorph rodents (electronic supplementary material, figure S1b and table S5) meaning that the absolute olfactory bulb size may have increased through time in rodents.
Paramys delicatus (AMNH 12506) and Pa. copei (AMNH 4756) have a longer circular fissure (figures 2 and 4; electronic supplementary material, figures S3 and S4) compared with Is. typus (e.g. AMNH F: AM 144836) and S. carolinensis (AMNH 258346)  (electronic supplementary material, figure S5). This could be due to an expansion of the frontal lobe of the cerebrum onto the circular fissure through time in Ischyromyinae and Scuiridae. The cerebrum does not fully cover the midbrain in either specimen of Paramys (AMNH 12506; AMNH 4756), Is. typus (e.g. AMNH F: AM 144836, ROMV 1007), R. turpanensis, L. kayi, plesiadapiforms or other Tertiary rodents [6–11,19–24] (electronic supplementary material, figure S5). This suggests that exposure of the midbrain is likely to be the primitive condition for euarchontoglirans. The caudal colliculi (part of the midbrain; = inferior colliculi) are not visible in Pa. delicatus (AMNH 12506) and Pa. copei (AMNH 4756), similar to some specimens of both Is. typus (ROMV 1007, AMNH 12252) and M. annectens (UW 12362) but are observable in other specimens of Is. typus (e.g. AMNH F: AM 144836), in a number of plesiadapiform primates (Plesiadapis cookei; Plesiadapis tricuspidens; M. cf. elegans; M. annectens, UW 14559; Ignacius graybullianus), a primitive apatemyid (L. kayi) and several Tertiary rodents (see figures 2a, 3a and 4 and also [6–11,23,24]; electronic supplementary material, figure S5). Although, the exposure of the caudal colliculi may be a primitive trait, it could alternatively be a derived feature related to sensory specialization [6,9,27,28].
Neocortical sulci are common features of brains that exceed 5 g . Since both species of Paramys would have had brains exceeding this mass, infolding of the neocortex may be expected . However, both Paramys endocasts are lissencephalic except for a weakly developed lateral sulcus (figures 2a and 3a). Despite their inferred higher brain mass, Pa. delicatus (AMNH 12506) and Pa. copei (AMNH 4756) do not exhibit more sulci than Is. typus (e.g. 5.5 cm3, ROMV 1007) or M. annectens (5.9 cm3, UW 12362) [6,9].
The cerebrum does not cover the cerebellum in Pa. delicatus (AMNH 12506) or Pa. copei (AMNH 4756) and the vermis is clearly separated from the two lateral lobes by the paramedian fissures (figures 2a and 3a). Both paraflocculi are well preserved in both specimens (electronic supplementary material, figures S3 and S4). Both specimens of Paramys (1.03%, AMNH 12506; 1.20%, AMNH 4756) have a smaller relative parafloccular volume compared with Is. typus (e.g. 1.60%, AMNH F: AM 144836; table 1) and S. carolinensis (2.03%, AMNH 258346; electronic supplementary material, table S1). This could suggest an enlargement of the paraflocculi relative to the volume of the rest of the brain through time in Rodentia. As the paraflocculi play a role in eye movement control , this function may have been enhanced in modern sciurids. No relevant comparative data have been published for other euarchontoglirans.
Various casts of the openings for the cranial nerves can be seen on the ventral surface of both Pa. copei and Pa. delicatus (figures 2b and 3b). The hypophyseal fossa for the pituitary gland is positioned at the level of the mandibular nerve (V3) in Pa. delicatus (AMNH 12506), Pa. copei (AMNH 4756) (figures 2b and 3b) and Is. typus (AMNH F: AM 144836) (, fig. 5b). This contrasts with the situation in S. carolinensis (AMNH 258346) in which the fossa is positioned more anteriorly relative to V3 (, fig. 6b) (electronic supplementary material, figure S5). The foramen rotundum is confluent with the sphenorbital fissure in most rodents ; this condition is primitive for eutherians  and is exhibited by many other mammalian orders (e.g. Dermoptera, Chiroptera, Carnivora) and is also true of Is. typus (, figs. 2b and 5b), Pa. delicatus (AMNH 12506) and Pa. copei (AMNH 4756) (figures 2b and 3b). In the extant rodent S. carolinensis, the foramen rotundum and the sphenorbital fissure are distally separated and then fused rostrally (, fig. 6b). In Primates, the foramen rotundum can be separated (M. annectens) or fused with the sphenorbital fissure (Ig. graybullianus, Pl. tricuspidens) [8–10]. Wahlert  described two foramina exhibited by Pa. copei and Pa. delicatus that transmit two branches of the maxillary nerve (V2): the masseteric and the buccinator nerves. The casts of those two foramina are observable on both endocasts of Pa. copei and Pa. delicatus (AMNH 4756, AMNH12506; figures 2b and 3b). In S. carolinensis, those foramina are united into a single foramen (, fig. 6b), and there is no corresponding foramen in Is. typus (, figs. 2b and 5b). The cast of one hypoglossal foramen is present in AMNH 12506 and AMNH 4756 , which is similar to S. carolinensis and other early Tertiary euarchontoglirans [6,8–12,31,33]. This contrasts with the two openings observed in the other ischyromyid, Is. typus [6,31].
The intracranial dural sinus system of Pa. delicatus and Pa. copei is typical of therian mammals . The superior sagittal sinus is visible and continuous with the transverse and sigmoid sinuses, which then connects with the jugular foramen in Pa. delicatus, Pa. copei, Is. typus and S. carolinensis, plesiadapiforms and apatemyids [6,8–11] (electronic supplementary material, figure S5). In Pa. delicatus, Pa. copei and Ig. graybullianus, the sigmoid sinus is also continuous with the condyloid vessels  (figures 2b and 3b). The superior sagittal sinus is pinched in appearance near its centre in Pa. copei (figure 2a), similar to what is observed in the caudal part of the superior sagittal sinus of S. carolinensis . This suggests that this sinus would have been located deep in the meninges . The jugular foramen is connected with the cast of the inferior petrosal sinus in Pa. delicatus, Pa. copei (figures 2b and 3b), Is. typus, S. carolinensis and M. annectens [6,9]. This sinus is not as well marked as in Ig. graybullianus  and is not preserved in Pl. tricuspidens or L. kayi [10,11].
According to Wible , the alisphenoid canal generally transmits the ramus infraorbitalis of the stapedial artery into the cranial cavity towards the sphenorbital fissure in rodents. Wahlert  described an alisphenoid canal for Pa. delicatus (USNM 23556) and Pa. copei (AMNH 4756). Nevertheless, the cast of the alisphenoid canal is only visible in Pa. copei, in which it appears to merge with the sphenorbital fissure (figure 2b, AMNH 4756). The transverse canal transmits veins that connect the two internal maxillary veins . In Pa. copei (AMNH 4756), the cast of the transverse canal enters the sphenorbital fissure medially from the alisphenoid canal (figure 2b). The cast of the canal cannot be identified in Pa. delicatus (AMNH 12506) but its opening is visible on the surface of the cranium in USNM 23556 , so its absence on the endocast may be due to poor preservation in this specific area of AMNH 12506. The transverse canal is also preserved in Is. typus (ROMV 1007)  (electronic supplementary material, figure S5c). In Pa. copei (AMNH 4756) and Pa. delicatus (AMNH 12506), the stapedial and facial canals are unconnected, which differs from the condition for the extant S. carolinensis  and another fossil rodent, Sciuravus nitidus (AMNH 12531), discussed by Wahlert , in which those canals are connected for a limited distance. The cast of the stapedial artery is also identifiable in apatemyids (L. kayi and Carcinella sigei) [11,12]. With respect to plesiadapiforms, M. annectens (UW 12362) has a stapedial artery , unlike Ig. graybullianus (USNM 421608) in which this artery is absent . The presence of one or more rami temporales (branches of the stapedial artery) is primitive for eutherians and is exhibited by many mammalian orders . This trait can either be preserved in all the members of a specific order (e.g. Pholidota, Chiroptera) or lost secondarily in some groups (e.g. Rodentia, Primates, Scandentia and Carnivora) . Paramys delicatus (USNM 23556) and Pa. copei (AMNH 4756) both have two rami temporales (figures 2 and 3; electronic supplementary material, figures S3 and S4), which is similar to the situation reported by Wible et al.  for another fossil rodent, Exmus mini (Ctenodactyloidea). The course of the internal carotid artery is transpromontorial in Pa. copei (AMNH 4756) and Pa. delicatus (AMNH 12506), which corresponds to the primitive condition in mammals, leading to a situation in which the promontorial artery is visible on the surface of the endocast (figures 2b and 3b), as in S. nitidus (AMNH 12531), described by Wahlert . The virtual endocasts of L. kayi and M. annectens exhibit the same condition [9,11]. The cast of the internal carotid artery is visible in the endocast of Is. typus, but as in S. carolinensis, the promontorial artery is absent (electronic supplementary material, figure S5e) [36,37] (, figs. 2 and 5).
3. Brain size and encephalization quotient
The endocranial volumes of Pa. delicatus (AMNH 12506) and Pa. copei (AMNH 4756) are 12.6 cm3 and 7.5 cm3, respectively. In order to obtain the encephalization quotient (EQ), the cranial capacity or endocranial volume (mm3) was converted to brain mass (g) by dividing the endocranial volume by 1.05 . The published EQs of apatemyids, plesiadapiforms and fossil euprimates were originally calculated based on volume [7–12], so their EQs were recalculated to be comparable to those calculated for rodents. Skull length and area of the cheek–teeth row were used to estimate body mass of the fossils since they are the dimensions that give the best estimation for body mass in fossil rodents . The rodent-specific equation of Pilleri et al.  was used to compare the EQs of Pa. delicatus and Pa. copei with those of other rodents (table 1; electronic supplementary material, tables S3 and S6). Using an estimate of body mass based on cheek–teeth area, the EQ of Pa. delicatus is 0.75 and Pa. copei is 0.84. The Orellan Is. typus has an EQ in the range of the genus Paramys, with the lowest value exhibited by ROMV 1007 (0.51) and highest by AMNH F: AM 144638 (0.77). Both specimens of Paramys have a lower EQ compared with the extant S. carolinensis AMNH 258346 (1.55)  (figure 5a), and smaller brain mass than would be expected based on data for living sciuromorphs (electronic supplementary material, figure S1c and table S5). According to Jerison , many mammalian groups show an increase in relative brain size through time (e.g. Perissodactyla ; Primates (e.g. [9,42]); Artiodactyla ; Chiroptera ). Since Paramys is one of the oldest genera of Rodentia , a lower EQ compared to the Oligocene Ischyromys would have been expected, so the fact that Pa. delicatus has an EQ at the high end of the range for Is. typus, while the EQ of Pa. copei lies above that range, is contrary to expectations. Ecological factors may also have an impact on brain size variation in Rodentia [2,6,44]. According to Pilleri and co-workers, extant arboreal rodents have higher EQs compared with those of living terrestrial species . Paramys has been considered as an arboreal  or a scansorial animal  based on postcranial data but a terrestrial rodent based on cranial data . Considering these differing hypotheses, it is unclear if the high EQ of Paramys was linked to its locomotor habits. In any case, it is clear that rodents do not show a simple pattern of progressive brain size increase through time.
Comparisons were made to other early members of Euarchontoglires using a regression analysis of log brain mass versus log body mass (electronic supplementary material, figure S1c) and Eisenberg's equation  for calculating EQ. The Eocene Paramys has a higher brain mass than would be expected for an archaic mammal, but lower than expected for a fossil or extant euprimate although the residual for Pa. copei from the fossil euprimate line is very low (−0.086; electronic supplementary material, table S3). With respect to the EQ, Pa. copei (0.76) and Pa. delicatus (0.62) have higher EQs compared with the apatemyid L. kayi (0.41) but lower than a more recent member of the same family, C. sigei, with an EQ of 1.28 (electronic supplementary material, table S5) [11,12]. Both Paramys specimens (AMNH 1206 and AMNH 5647) have a higher EQ compared with both specimens of Plesiadapis (e.g. Pl. tricuspidens, 0.14) and other early Tertiary plesiadapiform primates M. annectens (UW 12362; EQ = 0.42) and Ig. graybullianus (USNM 421608; EQ = 0.61) (electronic supplementary material, table S5) [11–14]. Paramys delicatus (AMNH 1206) and Pa. copei (AMNH 5647) have an EQ in the range of some fossil euprimates such as the adapoid Smilodectes gracilis (0.60) and the omomyoid Notharctus tenebrosus (0.65) from the Middle Eocene . The EQs of extant Sciuromorpha overlaps the lower bound of extant Euprimates' range (figure 5a), but shows a marked degree of variation. Eocene rodents have a higher EQ compared with stem Primates but are in the range of Eocene Euprimates. Those results suggest that the EQs of early members of Euarchontoglires may have been more diversified than previously thought, and in particular that primates were not as markedly encephalized relative to other mammals from their time period as had been suggested [17,46,47].
4. Neocorticalization in Rodentia and Primates
The EQ is only a coarse indication of the relative size of the brain. A more meaningful way of looking at brain size variation through time is to investigate which parts of the brain may contribute to shifts in the overall relative size. The orbitotemporal canal is interpreted as representing a landmark for the rhinal fissure , corresponding to the separation between the palaeo- and the neocortex on an endocast. Although the relationship between these structures was initially established in primates , in Brauer & Schober , the rhinal fissure figured for Sciurus vulgaris is in the same position as the orbitotemporal canal on our endocast of S. carolinensis, suggesting that this relationship also holds in rodents. The position of this landmark provides information on the degree of development of the neocortex [40,49]. Jerison showed that relative neocortical surface area increased through time in Euprimates as well as in other mammalian groups such as Carnivora and Artiodactyla . More recently, Long et al.  suggested that since stem primates do not show clear neocortical expansion compared with other mammals from the Palaeocene and Eocene epochs, increase in neocortical surface area probably started at the base of Euprimates (figure 5b; electronic supplementary material, table S7). The orbitotemporal canal is positioned more dorsally in Pa. copei (AMNH 4756) and Pa. delicatus (AMNH 12506) compared with Is. typus (ROMV 1007) and S. carolinensis (AMNH 258346) (figure 6; electronic supplementary material, figures S3b and S4b). Both species of Paramys are also almost entirely lissencephalic. As a result, Pa. delicatus and Pa. copei possessed low neocortical surface ratios, in the range of archaic mammals, notably lower compared with the Early Oligocene Is. typus and extant S. carolinensis (figures 5b and 6; electronic supplementary material, table S7). This suggests an increase in relative neocortical area through time occurring in Rodentia. In the broader context of Euarchontoglires, Primates also exhibit neocorticalization through time, but seem to show differences in the timing and/or rate of change [40,49]. Indeed, Eocene Euprimates already show a more ventrally located orbitotemporal canal as well as a more expanded neocortex (e.g. 35.2%, Smilodectes gracilis) compared with stem Primates from the same epoch (e.g. 24.4%, Ig. graybullianus UF 26000)  (figure 6), to the Eocene age rodents documented here, or even to the Oligocene aged Is. typus. These results demonstrate that changes in EQ and neocortical surface area may not always follow the same pattern through time, and in particular suggest that the neocorticalization commenced earlier in Primates than in Rodentia.
The study of the virtual endocasts of Paramys provides the opportunity to look at the earliest stage of rodent brain evolution documented to date and offers new information on the primitive condition of the Euarchontoglires brain. Surprisingly, Eocene Paramys had an EQ in the range or even higher than the Oligocene Is. typus, which contrasts with the general trend for an increase in EQ through time proposed by Jerison . Neocortical expansion occurred through time in rodents but was less pronounced in this group compared with Primates during the Eocene and Oligocene epochs. The fact that Eocene rodents had a higher EQ but a lower neocortical ratio compared with contemporary stem Primates suggests that high EQ may not always be associated with expansion of the neocortex.
The article's supporting data can be found in the electronic supplementary material, S1, and the surface renderings of the endocasts have been deposited on the external repository Morphosource (http://morphosource.org).
O.C.B. and M.T.S. both contributed to conception and design of the study. O.C.B. acquired the CT data and F.A.-M. segmented out both specimens in order to obtain the endocasts. O.C.B. and M.T.S. carried out the analyses and interpretations of data, drafted the article and revised it critically for important intellectual content. All authors gave final approval for publication.
We have no competing interests.
This research was supported by an NSERC Discovery Grant and a UTSC Vice Principal Research Competitiveness Fund Grant to M.T.S., and Pilot Research Funding from the Department of Anthropology of the University of Toronto to O.C.B.
We thank J. Meng and R. O'Leary from the American Museum of Natural History (AMNH) for providing access to the specimens to be scanned. We also thank M. Hill from the AMNH Microscopy and imaging facility for scanning the specimens. We thank J. Meng and S. López-Torres for comments that improved the manuscript. We thank two anonymous reviewers for their feedback and suggestions.
- Received September 24, 2015.
- Accepted January 7, 2016.
- © 2016 The Author(s)
Published by the Royal Society. All rights reserved.