New quantitative data from a digital cranial endocast of Kryptobaatar dashzevegi (Multituberculata, Mammalia) and a new estimate of body mass

Introduction

Overall brain volume and general shapes and volumes of some external features of the mammalian brain can be inferred from cranial endocasts. Endocranial volume (EV) is typically incorporated in encephalization quotient (EQ) studies comparing EV (as a proxy for overall brain size) to body size for closely related taxa (Jerison, 1973). Overall brain size is useful for estimating sizes of many constituent parts of the brain such as the medulla, diencephalon, hippocampus, and cerebellum (Jerison, 1973). However, brain size does not appear to universally scale with neuron density and neuronal connectivity (Deacon, 1990; Striedter, 2005). Several comparisons exist between relative brain size and ecological or behavioral traits, but these types of correlations should be viewed with caution (Deacon, 1990; Healy and Rowe, 2007). Furthermore, recent studies suggest that EQs are heavily influenced by phylogeny and that phylogenetic generalized least squares need to be considered to calculate the phylogenetic encephalization quotient, also known as PEQ (Ni et al., 2019). Additionally, an analysis by Smaers et al. (2021) suggests that brain and body sizes scale differently in different clades of mammals and that one universal calculation for EQ in vertebrates or mammals (e.g., Jerison, 1973; Eisenberg, 1981; Hurlburt, 1996) is not as meaningful as clade-specific EQs or PEQs. Similarly, the number of neurons, neuronal density, and the glia/neuron ratio do not scale universally with brain mass, but rather have clade-specific scaling factors (Herculano-Houzel, 2021). Nonetheless, general EQ values are still important for studying the evolution of brain size through time (Macrini, 2006; Finarelli and Flynn, 2007; Rowe et al., 2011; Benoit et al., 2023), especially when the relevant phylogeny is unresolved, preventing the calculation of PEQs.

Several natural endocasts of multituberculate mammals were described, and volumes were estimated for multiple taxa (Simpson, 1937; Kielan-Jaworowska, 1983, 1986; Kielan-Jaworowska et al., 1986; Krause and Kielan-Jaworowska, 1993; Kielan-Jaworowska and Lancaster, 2004). A natural cranial endocast was described (Kielan-Jaworowska and Lancaster, 2004) from arguably the best-known and most carefully studied species of multituberculate, Kryptobaatar dashzevegi, from the Late Cretaceous of Mongolia (Kielan-Jaworowska, 1970; Kielan-Jaworowska and Dashzeveg, 1978; Rougier et al., 1996; Wible and Rougier, 2000).

We extracted a digital cranial endocast from computed tomography (CT) imagery of a well-studied skull of K. dashzevegi to obtain accurate volumetric and linear measurements from the endocranial cavity of this taxon. The same skull was described in monographic detail from the exterior (Wible and Rougier, 2000), and the interior cranial anatomy is described in an unpublished Master’s thesis (Kik, 2002), but linear and volumetric measurements from the interior of the skull were not reported.

We also compare endocranial volume means for extant mammals taken from the literature obtained using various methods, with EVs measured from digital endocasts and EV estimates based on measures from exteriors of endocasts from the same individuals. Measurement of EV from CT data is shown to be more accurate and precise than using conventional methods of estimating EV from the endocranial cavity (Kubo et al., 2008) in the fields of anthropology and forensics (Stull et al., 2014). Our study further evaluates the differences in measurement of EV from CT data versus analog estimates of EV; while our general results are not surprising, the magnitude of discrepancy between methods is surprising (see below). Accurately and precisely measuring EV is an important component of determining EQ and thus is significant for studying the evolution of relative brain size.

Materials and methods

We extracted a digital endocast from a single skull of K. dashzevegi, which was collected from the Upper Cretaceous Djadokhta Formation of Ukhaa Tolgod, Mongolia (Dashzeveg et al., 1995; Wible and Rougier, 2000) by a joint Mongolian Academy of Sciences–American Museum of Natural History Expedition (PSS-MAE). The specimen (PSS-MAE 101) belongs to the collections of the Paleontological and Stratigraphic Section of the Institute of Geology of the Mongolian Academy of Sciences in Ulaanbaatar, Mongolia (PSS, GI PST), but is currently on long-term loan to the Division of Paleontology of the American Museum of Natural History (AMNH). The skull is nearly complete with relatively little distortion and measures 26.0 mm in length from the tip of the snout to the occiput (Wible and Rougier, 2000). The ontogeny of PSS-MAE 101 is not specifically discussed by Wible and Rougier (2000), but they do mention that the sphenoid complex is completely fused in this and other specimens of Kryptobaatar that they examined, suggesting that they are all skeletally mature adults.

This skull was scanned at the University of Texas High-Resolution X-ray Computed Tomography Facility in Austin, Texas, USA (UTCT). For an explanation of high-resolution X-ray computed tomography technology, see Rowe et al. (1995, 1999), Carlson et al. (2003), and the UTCT website (https://www.ctlab.geo.utexas.edu/about-ct/). The skull was scanned in its entirety at 140 kV in the coronal plane, resulting in 364 digital “slices” with dimensions of 512 × 512 pixels. The slice thickness is 0.11 mm, and an interslice spacing (the space between consecutive slices) is 0.08 mm for an overlap of 0.03 mm between consecutive slices. The field of reconstruction (FR) for the two-dimensional slices is 21.5 mm. The voxel dimensions in the X and Y planes for this scan were calculated by dividing the FR by the pixel size, yielding 0.04199 mm/pixel.

An endocast was digitally isolated from the skull using the program VGStudioMax^© and following the protocols described by Macrini et al. (2006, 2007a). Videos of the CT slices and digital endocasts discussed in this paper are available on the DigiMorph website.

Results

Because a natural endocast of K. dashzevegi (GI PST 8-2) was previously described (Kielan-Jaworowska and Lancaster, 2004), a detailed description of the digital endocast is not presented here. However, we do comment on certain aspects of the endocranial anatomy of Kryptobaatar.

The rhinal fissure did not leave a distinctive mark on the digital endocast of Kryptobaatar or on any other multituberculate endocast described so far (Kielan-Jaworowska, 1983, 1986; Kielan-Jaworowska and Lancaster, 2004). The rhinal fissure marks the histological division of the mammalian isocortex or neocortex from the piriform cortex (Nieuwenhuys et al., 1998) and can be used as an indicator of the expansion of this portion of the brain in fossil mammals if it is visible on the endocast. However, the rhinal fissure is often not visible on the endocasts of small mammals such as the extant marsupial Monodelphis domestica despite the structure being clearly visible on the corresponding brain (Macrini et al., 2007c). This is likely due to the thickness of the cranial meninges. The neocortex is present in extant monotremes and therians, and it is inferred to have been present in the last common ancestor of mammals (Nieuwenhuys et al., 1998; Ashwell, 2013); fossil evidence suggests that it arose in the mammalian stem-group by the origin of Mammaliaformes in the Late Triassic (Rowe et al., 2011). Therefore, based on extant phylogenetic bracketing, it is inferred that the brain of Kryptobaatar had a rhinal fissure and neocortex.

The pons of the hindbrain did not leave a noticeable indentation on the ventral surface of the endocast, and therefore, the previous claim that the trigeminal nerve inserted in the hindbrain anterior to the pons in multituberculates cannot be evaluated for K. dashzevegi (Kielan-Jaworowska, 1986). We labeled the dorsal protrusion on the cerebellum as the vermis (Figures 1A, B). This follows previous interpretations of this endocast structure being formed by the vermis (Simpson, 1937; Kielan-Jaworowska, 1983, 1986; Krause and Kielan-Jaworowska, 1993) but is contrary to others interpreting it to be the superior cistern (Kielan-Jaworowska and Lancaster, 2004). Kielan-Jaworowska and Lancaster (2004) pointed out the large superior cistern in the koala, Phascolarctos cinereus (Haight and Nelson, 1987), as being a structure that might obscure the vermis from view in an endocast, and applied this to their interpretation of the endocast of Kryptobaatar. Subsequent studies of the koala brain and comparisons between brains and endocasts of the same species of extant mammals show that the vermis does leave an impression on the corresponding endocast, even when the superior cistern is particularly large (Miguel and Henneberg, 1998; Taylor et al., 2006; Macrini et al., 2007c; Gilissen and Smith, 2012).

In general, the natural and digital endocasts of K. dashzevegi are very similar in morphology, but there are some noticeable differences. Some anatomical features such as the circular fissure, median sulcus, and superior sagittal sinus are more defined on the natural endocast (Kielan-Jaworowska and Lancaster, 2004) compared to the digital endocast (Figure 1). These discrepancies are attributed to differences in preservation quality between the two specimens and not to any artifact of the scanning process or digital extraction of the endocast.

The most striking differences between the natural and digital endocasts are in overall size estimates and dimensions (Table 1; Kielan-Jaworowska and Lancaster, 2004). The measured endocranial volume of the digital endocast of PSS-MAE 101 (0.34 mL; Table 1) differs dramatically from an estimate of the EV from the natural endocast of GI PST 8-2 (0.84 mL; Kielan-Jaworowska and Lancaster, 2004).

Table 1. Endocast measurements for Kryptobaatar dashzevegi (PSS-MAE 101).

^aMeasured following the procedure of Macrini et al. (2007a).

^bValue includes olfactory bulb casts.

^cCombined value for bilateral structures.

^dLength measured along the long axis of the cerebral hemisphere cast in dorsal view and width measured perpendicular to the long axis.

^eMeasured in lateral view.

^fMeasured in dorsal view.

Discussion

The discrepancy in EV values between the natural and digital endocast can be attributed to differences in size of the two individuals and in the techniques used to obtain the values. The skull of PSS-MAE 101 is approximately 76% the length of the skull of GI PST 8-2 (26 versus 34 mm, respectively; Wible and Rougier, 2000; Kielan-Jaworowska and Lancaster, 2004). The anteroposterior length of the endocast of PSS-MAE 101 was measured as 1.5 cm (Table 1), whereas the length of the endocast of GI PST 8-2 is reported as 1.7 cm by Kielan-Jaworowska and Lancaster (2004), which is approximately 13% longer than the other endocast. Examination of the occiputs of both skulls and associated postcranial material does not provide any obvious indicators of the ontogeny of the two specimens (Kielan-Jaworowska and Dashzeveg, 1978; Wible and Rougier, 2000). Other than ontogenetic differences, the size discrepancy between the two specimens might also reflect individual variation or sexual dimorphism.

The EV of GI PST 8-2 was estimated by Kielan-Jaworowska and Lancaster (2004) using the following equation, in which E = estimated EV, and l, h, and w represent the length, width, and height, respectively, of the endocast:

E= 0.25πlwh

The above equation represents the formula for calculating 25% of the volume of a cylinder. Olfactory bulb volume was estimated separately from the rest of the endocast, and then the two values were added together (Kielan-Jaworowska and Lancaster, 2004) to obtain the whole endocast estimate of 0.84 mL.

As mentioned above, the EV of PSS-MAE 101 was measured from high-resolution CT imagery, whereas we do not have a CT scan for GI PST 8-2. Macrini (2006) found that estimating EVs of extant mammals using the technique of Kielan-Jaworowska and Lancaster (2004) consistently overestimated the EVs compared with those measured from CT data of the same specimens.

The volume of the digital endocast (PSS-MAE 101) of K. dashzevegi was estimated using the partial-cylinder approach as 0.95 mL. The EV estimate for the natural endocast (GI PST 8-2) is 147% larger than the EV value measured for the digital endocast. Adjusting for the difference in size of the two specimens of Kryptobaatar, the estimated EV for the natural endocast (0.64 mL) is 53% greater than the EV measured for the digital endocast of PSS-MAE 101. This great a difference is not likely a result of individual variation based on comparisons with extant mammals (Eisenberg and Wilson, 1981; Gittleman, 1986; Ashwell, 2008), nor is the discrepancy likely a result of ontogenetic variation based on extant samples (e.g., Macrini et al., 2007c).

Discrepancies in analog estimates versus measured EV have important implications for calculation of the EQ of Kryptobaatar and for other Mesozoic mammals. Body mass for PSS-MAE 101 was estimated using 14 different metrics from dentition and the humerus based on ordinary least squares linear regressions from an analysis of body masses in small mammals (Huang et al., 2024). Body mass estimates range from 26.807 to 96.595 g with a mean of 55.013 g (Table 2). The average mass estimate for PSS-MAE 101 compares favorably with the body mass estimate of 62.9 g for the clearly larger specimen of GI PST 8-2.

Table 2. Body mass estimates for Kryptobaatar based on ordinary least squares linear regressions from an analysis of body masses in small mammals (Huang et al., 2024).

Measurements of PSS-MAE 101 were taken from the data reported by Wible and Rougier (2000) and measured from a photograph of the humerus in the same source. Variables are based on the measurements defined by Huang et al. (2024).

h.length, humerus length; M1_l, first upper molar length; M2_l, second upper molar length; m1_l, first lower molar length; m2_l, second lower molar length; P1_l, first upper premolar length; P1_w, first upper premolar width; P2_l, second upper premolar length; P2_w, second upper premolar width; P3_l, third upper premolar length; P3_w, third upper premolar width; P4_l, fourth upper premolar length; P4_w, fourth upper premolar width; p4_l, fourth lower premolar length.

Using the measured EV of the digital endocast and the above-mentioned estimates for body mass, the EQ of Kryptobaatar ranges from 0.13 to 0.54 (Table 3). We used the average of the 14 body mass estimates (55.013 g), the measured EV, and the EQ equation of Eisenberg (1981) to calculate an EQ of 0.32 for PSS-MAE 101. An EQ of 0.71 is obtained by using the equation of Eisenberg (1981) and the EV estimate of the natural endocast from Kielan-Jaworowska and Lancaster (2004). A measured EV from the natural endocast would likely yield an EQ closer to 0.32 for that specimen.

Table 3. Encephalization quotient values calculated using the digital endocast of Kryptobaatar dashzevegi (PSS-MAE 101) based on body mass estimates from Table 2, the endocranial volume (with olfactory bulbs) of 0.34287 mL, and the equations from Jerison (1973) and Eisenberg (1981).

An EQ of 0.32 is significantly lower than those estimated for two other multituberculates [Ptilodus (EQ = 0.62) and Chulsanbaatar (EQ = 0.54–0.56); Kielan-Jaworowska, 1983; Krause and Kielan-Jaworowska, 1993; Kielan-Jaworowska and Lancaster, 2004]. For both Chulsanbaatar and Ptilodus, complete endocasts were not available to measure EV, and instead, models of the endocasts were constructed based on the exposed dorsal surfaces of endocasts and their EVs measured using volume displacement in water (Kielan-Jaworowska, 1983; Krause and Kielan-Jaworowska, 1993). It is unclear how the sides and ventral surfaces of the endocasts were reconstructed in these models. Therefore, it is likely that the EQs reported above for Chulsanbaatar and Ptilodus are also inflated because the estimates of EVs were estimated rather than measured from CT images, and it is difficult to directly compare those values with those of Kryptobaatar.

Therefore, we can compare the EQ of Kryptobaatar (0.32) with EQs calculated using the equation of Eisenberg (1981) for other key fossil mammals and stem mammals that have EV values measured from CT data. The stem mammal Hadrocodium wui has an EQ well above that of Kryptobaatar at 0.49, whereas the EQ of Morganucodon sp. is identical at 0.32 (Rowe et al., 2011). The stem therian Vincelestes neuquenianus has an EQ above that of Kryptobaatar at 0.37 (Macrini et al., 2007b), whereas the stem marsupial Pucadelphys andinus has an identical EQ of 0.32 (Macrini et al., 2007a). This suggests that the EQs of Kryptobaatar and multituberculates are significantly lower than previously estimated and also clearly dependent on the estimate of body mass (Table 3). If the EQ of Hadrocodium is taken as a proxy for the EQ for the ancestral crown mammal (Rowe et al., 2011), then Kryptobaatar has a significantly lower EQ than the presumed ancestral baseline for Mammalia.

Body mass estimates for extinct taxa also introduce a significant amount of variation (and error) when calculating EQs (see Table 2). Often, dental measures are used to estimate body mass in fossil mammals mainly because dentition preserves better in the fossil record than the rest of the skeleton. However, when postcranial elements are available, particularly the femur or humerus, estimates of body mass using those elements outperform estimates from dentition (Huang et al., 2024).

The lack of a clearly defined phylogenetic position for multituberculates within Mammalia and the lack of agreement on its precise taxonomic composition raise questions about whether or not Multituberculata is monophyletic (Kielan-Jaworowska and Hurum, 2001), both of which hamper the ability to calculate PEQs for the group (Ni et al., 2019). Therefore, at the present time, using the generalized equations of Jerison (1973) and Eisenberg (1981) for the calculation of multituberculate EQs is the only option. In the future, we hope that these phylogenetic questions can be resolved, and PEQs can be calculated for this group. Regardless, the EQ of Kryptobaatar is clearly lower than those previously reported for multituberculates in the literature, suggesting a dip in relative brain size for that group compared to that inferred for the base of crown Mammalia (Rowe et al., 2011). It is clear that methodological discrepancies and shortcomings with older estimates of EV using analog techniques, along with inconsistent estimates of body mass and the lack of accounting for phylogenetic effects on EQ (e.g., not using PEQs), warrant a reexamination of relative brain sizes for Mesozoic mammals and their closest extinct relatives.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Julien Benoit, University of the Witwatersrand, South Africa

Reviewed by: Emmanuel Paul Gilissen, Royal Museum for Central Africa, Belgium

Leonardo Kerber, Federal University of Santa Maria, Brazil