The body armor of ankylosaurians is a unique morphological feature among dinosaurs. Despite being studied for decades, paleohistological analyses have only started to uncover the details of its function. Yet, there has been an overall bias toward sampling ankylosaurian remains from the Northern Hemisphere and limited quantitative studies on the morphological and functional evolution. Here, we describe new ankylosaurian materials recovered from the Late Cretaceous of Antarctica that, in combination with data compiled from the literature, reveal new insights into the evolution of the ankylosaurian body armor. Based on histological microstructure and phylogenetic results, the new Antarctic material can be assigned to Nodosauridae. This group shares the absence/poor development of the basal cortex and highly ordered sets of orthogonal structural fibers in the superficial cortex. Our morphospace analyses indicate that large morphological diversity is observed among both nodosaurids and ankylosaurids, but they became more functionally specialized in late-diverging nodosaurids. Besides acting as effective protection against predation, osteoderms also exhibit highly ordered structural fibers in nodosaurids, enabling a decrease in cortical bone thickness (as in titanosaurs), which could have been co-opted for secondary functions, such as calcium remobilization for physiological balance. The latter may have played a key role in nodosaurid colonization of high-latitude environments, such as Antarctica and the Arctic Circle.

Material 

The figures of specimens CAV-4 and 5 before the paleohistological sections are available in Supplementary Fig. 1. There is no figure for the specimen CAV-10.

Statistical Analyses

The dataset compiled from the literature for our linear morphometric analyses is available in the file “Dataset_osteoderms.xlsx”. The compiled specimens are listed in Table 1. The entire dataset of PAST files to osteoderms is available in “Dataset_osteoderms.dat”. 

The abbreviations and description of variables used in both datasets are: B, thickness of the osteoderm basal cortex; BI, basal index (BI = 100B/T); CI, cortical index (CI = 100*(S + B)/T); Co, thickness of the osteoderm core; CoI, core index (CoI = 100Co/T); Group, main group that the specimens are assigned (Ankylosauridae, Nodosauridae or Titanosauria); Osteoderm, shape of the osteoderm; Group+osteodem, classification joining main group and shape of osteoderm; S, thickness of the osteoderm superficial cortex; S+B, sum of osteoderm superficial and basal cortex (=cortical thickness); Specimen, collection number of the analysed specimens; SI, superficial index (SI = 100S/T; T, total thickness of the osteoderm; Taxon, specific classification of the analysed specimens; all Log. variables are at the base Log10. These variables comprise linear measurements and relative thickness of osteoderm paleohistological slides (all in mm), comprising the compact bone walls (basal and superficial) and the core. They followed the proposal of Burns and Currie (2014).

We performed the Shapiro-Wilk (W) test for normality to evaluate if our data follows a normal distribution. We also performed the linear discriminant analysis (LDA) to discriminate the morphospace by reducing the dimension of the parameters.

Phylogenetic Analysis

The codification of the specimens based on the compilation of the matrix of Soto-Acuña et al. (2021)—the modification of the authors on the dataset of Arbour and Curie (2016). The matrix is available in the files “PhyloAnk.nex” and “PhyloAnk.tnt”. The description of the characters is available in “PhyloAnk.nex”.