Phylogenetic analyses were run on two established datasets used to explore rela- tionships of the Dromaeosauridae and Theropoda. The first phylogenetic analysis was run with 34 operational taxonomic units (33 ingroup OTUs and 116 charac- ters). These data were based on the study of Bell and Currie (2016), which was, in turn, based off the studies of Evans et al. (2013) and Longrich and Currie (2009). Data were run with TNT version 1.5 (Goloboff et al., 2016). This analysis results in 28,053 most parsimonious trees, each with a tree length of 241 steps, a Consis- tency Index of 0.527 and a Retention Index of 0.685 (Fig. 3.3). The strict consensus tree recovered the Eudromaeosauria in a large, unresolved polytomy with three small clades within, including Bambiraptor feinbergi + Saurornitholestes langstoni,
Deinonychus antirrhopus+Atrociraptor marshalli, andAdasaurus mongoliensis+Tsaa-
gan mangas(see supplemental info). All other eudromaeosaurs, includingDineobel-
lator notohesperis, were part of the polytomy. Previously recovered clades of South
American dromaeosaurids (Unenlagiinae) + Rahonavis from Madagascar and the microraptorines (Microraptorinae) were recovered as monophyletic. A majority rule consensus tree (set at 50%) provided more resolution with intrafamilial rela- tionships (Fig. 3.3). The Velociraptorinae was recovered with mostly Asian taxa, althoughAcheroraptorandBoreonykusform a subclade withVelociraptor mongolien- sis. The Dromaeosaurinae was recovered with two Asian taxa (Achillobator giganti- cusand the unnamed Bayanshiree Formation dromaeosaurid) andUtahraptorand
Dakotaraptor as progressively more basal. Deinonychus + Atrociraptor and Bambi-
raptor+Saurornitholestes langstoniform more basal clades, with Dineobellator noto-
hesperislying between them. The Microraptorinae was recovered as monophyletic
toMicroraptor. Dromaeosaurids from Gondwana are monophyletic, with the Ra-
honavisfrom Madagascar sister to South American taxa. Mahakala omnogovaewas
recovered as the sister taxon to all other dromaeosaurids.Dineobellator notohesperis
was also run through an established theropod dataset to further gather insight into its phylogenetic placement. The Theropod Working Group dataset was also used, mainly from Brusatte et al. (2014) and recently updated by Cau et al. (2015, 2017). While many theropod groups had higher resolution, intrafamilial relationships of the Dromaeosauridae were poorly resolved. The strict consensus majority rule tree for this dataset can be found with the supplemental information.
Maximum parsimony was used with the current dataset as this algorithm does not require prior assumptions of the data. It is also the one most often used by previous authors (e.g., Gauthier, 1986; Calvo et al., 2004; Currie and Varricchio, 2004; Makovicky et al., 2005; Csiki et al., 2010; Zheng et al., 2010; Senter et al., 2012; Turner et al., 2012; Brusatte et al., 2013; Evans et al., 2013; Han et al., 2014; Laco- vara et al., 2014; Sues and Averianov, 2014; DePalma et al., 2015; Lü and Brusatte, 2015; Bell and Currie, 2016; Gianechini et al., 2018), and allows for comparison of results between the current study and previous studies. While a few studies on di- nosaur phylogenies have utilized other methods, namely Bayesian inference (e.g., Prieto-Marquez and Wagner, 2009; Prieto-Marquez, 2010; Brusatte and Carr, 2016; Pei et al., 2017), these methods require prior assumptions and probabilities of the data. Maximum parsimony works well with datasets containing taxonomically- constrained OTUs, or members that are more closely related, such as those in- trafamilial. The current study works mainly within the family Dromaeosauridae, and the outgroup taxa are all part of the group Eumaniraptora. The more con- stricted taxonomic grouping being investigated makes maximum parsimony work
better than if the dataset included a much wider taxonomic range of OTUs. Max- imum parsimony can also have problems dealing with taxa that are drastically different ages close to each other. ForDineobellator notohesperis this would not be an issue as the other OTUs in its clade all occur in a relatively short time period (Campanian-Maastrichtian). This could have an effect on the clade includingDro-
maeosaurus albertensis, (Achillobator giganticus, the unnamed Bayanshiree Forma-
tion dromaeosaurid, and Utahraptor, as the latter taxon is significantly older than the others. This could also have an effect on the microraptorines as Hesperony-
chus elizabethaeis significantly younger than the other members of the clade. Using
maximum likelihood or Bayesian inference would likely have an effect on the posi- tion of these taxa, particularlyHesperonychus elizabethaeand Utahraptor. However, the monophyly of the Microraptorinae, with respect to Hesperonychus elizabethae
has not been questioned and, although its position within the clade may vary, its phylogenetic position as a microraptorine is not likely to change, even due to the significant ghost lineage present between it and other Early Cretaceous microrap- torines. Although the ghost lineage is not as long between Utahraptor and other members of its clade, there is a higher possibility that its position as a member of the clade would not be retained in a non-maximum parsimony phylogenetic analysis. It is noted that Brusatte and Carr (2016) reached a similar conclusion when looking at tyrannosaurids. While maximum parsimony and Bayesian anal- yses resulted in similar phylogenetic trees, they hypothesized the differences that were present were due to a large gap in the fossil record of the group with no taxa present, causing slight discrepancies between their two analyses. Based on their results,Utahraptormight be pulled from its current clade, and would likely instead be sister to the clade. However, even if that was the case, it would not be deter- minedly more correct than what is shown in the maximum parsimony analysis in
the present study. The only way to be more certain of the results will be to fill in the gap or hiatus in the dromaeosaurid fossil record between Barremian and Cenoma- nian. Until that is accomplished, the current position ofUtahraptor is considered as accurate as current data allows. The position ofDineobellator notohesperiswould also not be significantly altered regardless of the algorithm used to produce the phylogenetic tree, and its major relationships would remain consistent.