MATERIALS AND METHODS

The Specimens

Locality Data. F.AM 42891 (specimen from the Frick American mammals collection at the AMNH, hereafter referred to as the AMNH skull) is an isolated skull from the Whitneyan of Shannon County, South Dakota. Data from the original specimen label cites Skinner and Mefferd as the collectors from northeast of Indian Stronghold on the divide between west Big Corral Draw and Cottonwood Creek from a lower Protoceras channel. The label mentions that this would be in the upper Oreodon beds or lower Poleslide Member of the Brule Formation of Jim Bump.

SDSM 59566 (hereafter referred to as the SDSM skull) is an isolated skull from the Orellan of South Dakota, collected by Japeth Boyce, who donated it to the South Dakota School of Mines, where it has been on display in the Museum of Geology. Japeth Boyce provided the following provenance information (Boyce, personal commun., 2005): specimen collected from the ‘Metamynodon sandstone,’ a channel sand in the lowest Brule Formation located just below the lower nodular layer; site approximately midway between Cottonwood Pass and Stronghold Table.

Preservation and Relative Maturity. Although both are partially crushed and distorted, the two skulls described here are generally well preserved and relatively complete. The AMNH skull is horizontally flattened (Figure 1A), and the SDSM skull is compressed sagittally and sheared (Figure 1B), with the right side elevated relative to the left. The different compaction planes of these two specimens present somewhat complementary information for interpreting the degree of anatomical deformation, but precise determination of the true skull shape is not possible. Thus, for example, while the AMNH skull suggests a wide rostrum for Colodon, the SDSM rostrum is narrow, and the true shape is some intermediate condition.

The braincase of the SDSM skull is more completely preserved than the AMNH skull, which lacks posterior parietals and most of its occipital shield, including the basioccipital and supraoccipitals. The critical narial region, which is the focus of this description, is better preserved on the AMNH skull. The SDNH skull only preserves fragments of its nasals, and most of the narial opening margins are damaged.

Based on dental eruption and suture closures, the AMNH skull is less mature than the SDSM specimen, and neither represents an adult condition. The immaturity of the AMNH skull is indicated by the incomplete eruption of its permanent premolars and molars; M3 is still within its crypt, and it retains deciduous premolars. The dental formula differs on the left and right sides, most likely a consequence of dP3-4 having been prepared away on the left side. The dental stage is as follows: right side, P1 missing (adult roots intact), P2, P3-4 in crypts, M1-2, M3 in crypt; left side, P1-2, dP3-4 (P3-4 in crypts), M1-2, M3 in crypt (Figure 2). This arrangement is consistent with an eruption sequence pattern observed in recent tapirs (Colbert 1999). The presence of incisors is inferred from alveoli, but canines are lacking. The AMNH specimen’s immaturity is also indicated by the largely open cranial sutures. For example, the basioccipital is missing, having separated along the open spheno-occipital synchondrosis.

All the adult premolars and M1-2 are in place on the SDSM skull, but M3 lies within its crypt. I1-2 are preserved on the left premaxilla, and I1 and I3 on the right premaxilla (Figure 3). As on the AMNH skull, canines are lacking. Additional indicators of the immaturity of this specimen are the lack of fusion of the exoccipitals to the supraoccipitals, and of the basioccipital to the basisphenoid.

The Referral of These Specimens to Colodon. The monophyly of Colodon is suspect, as indicated by a series of phylogenetic analyses performed by Colbert (1999). This condition is largely a consequence of the limited material, particularly nondental material that has been referred to Colodon. Traditionally, Colodon has been identified by its distinctive teeth (e.g., Colbert and Schoch 1998; Radinsky 1963). Radinsky (1963) recognized three North American species of Colodon (C. kayi, C. woodi, and C. occidentalis).

Dental characters are the basis for the referral of the skulls described here to Colodon. These similarities include both the degree of molarization of their premolars, and their lingually displaced and reduced molar metacones (Figure 1, Figure 2; Radinsky 1963). The AMNH skull cannot be unquestionably referred to any particular species of Colodon, but the SDSM skull falls comfortably within the range of C. occidentalis. Species of Colodon were discriminated by Radinsky (1963) based on size differences in their teeth. The dentition of the AMNH skull was compared favorably to Colodon occidentalis by Radinsky (1963) who noted, however, that it was generally larger than other C. occidentalis, and would also represent a chronostratigraphic range extension from the Chadronian and Orellan into the Whitneyan North American Land Mammal ‘Age’. Radinsky (1963) suggested that this specimen might record an evolutionary increase in size within the C. occidentalis lineage over time.

The slightly smaller dentition of the Orellan SDSM skull falls within the size range for C. occidentalis, to which it is referred (compare measurements presented in Radinsky 1963 and dental measurements for the two skulls provided in Table 1). Although it cannot be assumed that the two Colodon skulls described here belong to a single species, it is clear that they were closely related based on overall morphological similarities.

HRXCT Scanning and Image Processing

Both Colodon skulls were scanned at the University of Texas High-Resolution X-ray CT Facility (UTCT) using the high-energy subsystem as described by Ketcham and Carlson (2001). The original data sets for both as are saved as1024 by 1024 pixel TIFF images with a 16-bit gray scale depth. The slice thickness for the AMNH skull was 0.50 mm, and the inter-slice spacing 0.40 mm. The field of reconstruction was 125 mm, yielding an in-plane resolution of 0.122 mm/pixel and 556 slices in the original coronal plane. An animation of the original CT data of the AMNH skull, reduced from their original image size for web viewing, can be seen in Appendix 1. The original data set for the SDSM skull comprises 576 slices in the coronal plane, each slice having a thickness of 0.50 mm and an inter-slice spacing of 0.45 mm. The field of reconstruction was 112 mm, yielding an in-plane resolution of 0.109 mm per pixel. >An animation of the original CT data of the SDSM skull, reduced from their original image size for web viewing, can be seen in Appendix 2.

Animated three-dimensional (3D) renderings, and ‘dynamic cutaway’ views of the skulls (see DigiMorph: Colodon AMNH and DigiMorph: Colodon SDSM) were produced using VGStudioMax software.

Comparative CT Data and Terminology

The descriptions presented here make reference to an HRXCT dataset for Protapirus (SDSM 2829; see DigiMorph: Protapirus), a White River contemporary of Colodon, whose cranial anatomy is generally much less derived than Tapirus. The original data sets for Protapirus are saved as1024 by 1024 pixel TIFF images with a 16-bit gray scale depth. The slice thickness was 0.50 mm, and the inter-slice spacing 0.50 mm. The field of reconstruction was 140 mm, yielding an in-plane resolution of 0.1637 mm/pixel and 602 slices in the original coronal plane. An animation of the original CT data of Protapirus, reduced from their original image size for web viewing, can be seen in Appendix 3. Anatomical terminology generally follows Witmer et al. (1999).

Phylogenetic Analysis

The evolutionary position of Colodon within Ceratomorpha (exclusive of all but the most basal rhinocerotoids) was estimated using character data derived from both literature descriptions and observed specimens (Appendix 4, Appendix 5). The analyzed data matrix includes 22 in-group taxa and 89 cranial and dental characters (Appendix 4, Appendix 5) that were largely developed from the literature. The matrix includes specimens referred to a ‘new San Diego taxon,’ described by Colbert (1999). The data matrix (Appendix 5) was analyzed with PAUP v4.08b (Swofford 2003), using a heuristic search option, characters unordered, and keeping only the most parsimonious trees. Outgroups include Homogalax protapirinus, Isectolophus latidens, and I. annectens. Diagnoses assume ‘Deltrans’ character optimization.