The Proceedings of the Eighth International Conference on Creationism (2018)

that some of the taxa with less than 50% of available character data are transitional in morphology between the other groups so they fall in between them, creating a spatially-scattered ordination for the plot. This may be occurring in some situations, but removal of taxa with less than 50% of characters across the board should not preferentially remove taxa with transitional morphologies. As a result, we predict that future datasets with more complete data will continue to reveal these three major dinosaur divisions as discontinuity-bounded, morphospatially separate associations (apobaramins). B. PCA-defined Holobaramins Many dinosaur holobaramins could be clearly distinguished. Often, the additional material of newer datasets clarified uncertainty from earlier, less complete matrices (i.e., 2004 matrices). One example was “Prosauropoda.” The 2004 BDC results (Fig. 42) showed positive BDC between all non-sauropod sauropodomorphs (except for Camelotia , which was only correlated with a few others). Both MDS and PCA suggested there may be discontinuities within this group, but it was not very clear. However, the Bronzati (2017) PCA results show various clusters and series of “prosauropods”, which probably correspond to different holobaramins such as Thecodontosauridae and non-sauropod Massopoda (listed as monobaramins (but possibly holobaramins) in Table 1 and Appendix Table 2. Another group that benefited from better data was theAnkylosauria. Ankylosauria BDC results for the 2004 matrix separated taxa into two groups. One group contained traditional nodosaurids and two polacanthines (a group of ankylosaurs that have been taxonomically difficult to place as either ankylsoaurids or nodosaurids). The non- polacanthine nodosaurid block was separated from the ankylosaurid block by negative correlation, but the polacanthines shared no correlation with the ankylosaurids at all. This pattern was explained well by the MDS plot and especially the PCA plot, which both show the polacanthines in a gap directly between the ankylosaurids and nodosaurids. These patterns suggest that all of Ankylosauria might be a holobaramin. Interestingly, the results from the Zheng et al. (2018) analysis show two different patterns. Comparing PC 1 and PC 2 results in a series of nodosaurids and a separate cluster of ankylosaurids. Polacanthine taxa fall in between. Comparing PC 1 and PC 3, however, shows the polacanthines at the base of a series of nodosaurids, far away from the ankylosaurids by PC 3, but still near them in PC 1. Surprisingly, the ankylosaurid Crichtonpelta clusters near the polacanthines and not the other ankylosaurids. These PCA results for Zheng et al. (2018) help to show why it is so difficult to classify polacanthines: some of their features are more ankylosaurid-like and some are more nodosaurid-like. It is likely, given these results, that ankylosaurs as a whole are a single holobaramin. Indeed, recent research into the evolution of the tail club in ankylosaurids (Arbour and Currie 2015) is consistent with the idea that nodosaurids and ankylosaurids share a single common ancestor. Some of the taxa that are considered transitional in tail morphology between the two groups have not been included in this analysis ( Liaoningosaurus and Gobisaurus ). Hadrosauridae provided an example of group with potentially hidden morphological divisions. PCA supported the groupings of BDC and MDS (compare Figs. 78-80) yet PCA revealed a potentially deeper subfamily-level division. Hadrosauridae had positive BDC between group members, although two blocks of positive correlation within the hadrosaurid block were evident, Doran et al. ◀ Dinosaur baraminology ▶ 2018 ICC 444 Figure 96. Biplot of PCAscores (black) and vectors (gray) for Fry’s (2015) data matrix for Ceratopsidae. PC 1 accounts for 45.6% of the variance and PC 2 accounts for 19.2% of the variance. PC 1 separates Chasmosaurinae from non-chasmosaurine taxa. PC 2 reveals a separation between Triceratopsini (top) and non-triceratopsin chasmosaurines (below). Figure 97. Biplot of PCAscores (black) and vectors (gray) for Fry’s (2015) data matrix for Ceratopsidae. PC 1 accounts for 45.6% of the variance and PC 3 accounts for 14.0% of the variance. PC 3 divides centrosaurine taxa (top) from Chasmosaurinae (below).