Braincase and digital endocast of a loricatan pseudosuchian (Reptilia: Archosauria) from the Upper Triassic of Nova Scotia (Canada)

We present a detailed description of an incomplete but excellently preserved braincase and a digitally generated endocast of a pseudosuchian archosaur from the Upper Triassic (Carnian) Evangeline Member of the Wolfville Formation of Nova Scotia (Canada). The general morphology of the braincase particularly resembles that of the rauisuchid Postosuchus kirkpatricki from the Upper Triassic (Norian) of Texas. Indeed, the quantitative phylogenetic analyses found the Wolfville specimen within Rauisuchidae (Postosuchus spp., Polonosuchus silesiacus, and Rauisuchus tiradentes) among loricatan pseudosuchians based on the presence of a deep and dorsoventrally elongate basisphenoid recess. Although the specimen described here shows differences with Postosuchus kirkpatricki and Postosuchus alisonae, its fragmentary condition and the absence of overlapping elements with other rauisuchids (Polonosuchus silesiacus and Rauisuchus tiradentes) led us to be cautious and identify it as aff. Postosuchus sp. The Wolfville rauisuchid braincase provides evidence for the presence of this clade in low paleolatitudes during the Carnian, bridging the higher paleolatitude occurrences of Rauisuchidae in southern Brazil and in Poland, respectively.

Methods
Specimen scan and digitalization

The surfaces of the braincase YPM VPPU 020750 were 3D-scanned with an Artec Spider 3D Scanner, and the scanning data were collected and processed in Geomagic Studio (3D systems, Rock Hill, South Carolina). Images of 3D surface meshes were produced using MeshLab.

Subsequently, the specimen was mCT-scanned using a GE Phoenix v|tome| xm 240/180 Dual mCT system at the National Museum of Natural History (Smithsonian Institution) with a voxel size of 125 mm, a source voltage of 175 kV, and a source current of 150 mA. It generated 1432 images, each with a height of 2024 pixels and a width of 2014 pixels. The scan data were exported in DICOM format and subsequently imported into the digital visualization software Avizo (v.9.1). The endocast was reconstructed using the Segmentation Editor with key anatomical structures isolated and highlighted in different colors. Finally, images of the digital endocast visualized inside the transparent braincase were exported.

Phylogenetic analyses

The phylogenetic relationships of YPM VPPU 020750 were tested using the character-taxon matrix of Desojo and Rauhut (2024), which is the latest iteration of the dataset originally published by Nesbitt (2011). After adding YPM VPPU 020750, the modified version of the data matrix is composed of 454 characters scored across 97 active terminals (Supplementary Material). The data set was analyzed under implied weighting maximum parsimony in the program TNT version 1.6 (Goloboff and Morales 2023). This decision of weighting against homoplasy follows the results of the analyses of Goloboff et al. (2018) (based on simulations) and Ezcurra (2024) (based on empirical data), in which implied weighting outperformed equal weighting in topological accuracy and stability, respectively. A series of analyses were conducted with a range of concavity constant values (k) between 5 and 11 following the suggestion of Ezcurra (2024) for a matrix with the number of terminals used here. The following 22 characters were considered as ordered following Desojo and Rauhut (2024): 32, 52, 121, 137, 139, 156, 168, 188, 223, 247, 258, 269, 271, 291, 297, 314, 328, 356, 371, 399, 413, and 446. The search strategy initially used a combination of the tree-search algorithms sectorial searches, drifting, ratchet, and tree fusing, until 100 hits of the same minimum tree length were achieved. The shortest trees obtained were then subjected to a final round of TBR branch swapping. Zero length branches in any of the recovered MPTs were collapsed. All the trees were rooted with Mesosuchus browni. Homoplasy indices for each analysis under the different k values were calculated with ‘STATSb.run’ (Spiekman et al., 2021). Group supports were quantified using no-zero weight symmetric resampling analyses, using 10,000 pseudo-replications (each with 10 replications of Wagner trees + TBR) and reporting both absolute and GC (group present/contradicted) frequencies. Finally, a global strict consensus tree (GSCT) was generated from all the most parsimonious trees (MPTs) found in all the analyses using the different k values. Similarly, absolute and GC resampling frequencies were calculated from all the resampling trees recovered using the different k values and plotted on the branches of the GSCT.

These analyses were implemented using the TNT script ‘phylo_analyses_IW_and_eqW.run’, which uses a series of arguments to customize automatized analyses (see README file in Supplementary Material). This script, ‘STATSb.run’, and data matrix files should be in the same folder (working directory). The script needs the following seven arguments that allow the user to customize the analysis: 1) the name of the matrix file without the “.tnt” extension, 2) character weighting treatment (0 = implied weighting, 1 = implied weighting and equal weights, 2 = equal weights), 3) the lower limit of the k values range, 4) the upper limit of the k values range, 5) type of tree search (0 = traditional search, 1 = new technology search, 2 = consensus stabilization search), 6) number of iterations during tree search, and 7) the number of pseudo-replications of the resampling analyses. Hence, to reproduce the analyses conducted here, the script should be run as follows in TNT: ‘run phylo_analyses_IW_and_eqW.run Nesbitt_matrix_YPM 0 5 11 1 100 1000;’.