Inferring the phylogenetic relationships among extant and fossil taxa is critical to understanding the evolutionary history of any group, and the study of human origins is no exception. However, the results of any phylogenetic analysis are highly dependent on both the character data and taxon sampling used. In particular, the choice of outgroup taxa can have a dramatic effect on the resulting cladograms (Nixon and Carpenter, 1993; Wilberg, 2015). The use of outgroups establishes character polarity by utilizing data from taxa that sit outside of the clade of interest (ideally the most closely related sister taxa) to determine which character states should be considered apomorphic (derived) or plesiomorphic (primitive; Maddison et al., 1984; Wiley, 1991; Bryant, 2001). In order to infer cladograms that most accurately reflect evolutionary relationships, it is important to select suitable outgroups that are likely to sample the ancestral states of the ingroup taxa, thereby allowing an accurate assessment of polarity for the analysis.

Previous work has considered the importance of fossil data in establishing character polarity in morphological analyses of hominins, other primates, and other vertebrates more broadly (Gauthier et al., 1988; Benefit and McCrossin, 1991; Leakey et al., 1991; Strait and Grine, 2004). Though extant taxa may be informative, particularly because they often offer a high level of character sampling, the use of extant outgroups limits the reconstruction of ancestral character states to only those states present in living organisms, which may not be representative of the full range of features that characterize a clade or may include convergent character states (Gauthier et al., 1988). In many cases, fossils present intermediate states that provide better resolution of an evolutionary transformation series than do extant taxa (e.g., Gauthier et al., 1988; Benefit and McCrossin, 1991). Therefore, the use of fossil taxa as outgroups may produce cladograms that are more accurate, despite incomplete preservation (Doyle and Donoghue, 1987; Gauthier et al., 1988; Huelsenbeck, 1991; Eernisse and Kluge, 1993; Yang et al., 1996; Sumrall, 1997; Smith, 1998; Springer et al., 2001; Strait and Grine, 2004; Mongiardino Koch et al., 2021).

Several recent studies of hominin phylogenetics have used Colobus guereza and mixed-species samples of Papio (Papio anubis and Papio ursinus) as outgroups (Strait and Grine, 2004; Mongle et al., 2019). In these studies, cercopithecoids were selected as sister taxa to an ingroup containing all extant hominoids. The selection of C. guereza and Papio spp. was based upon the reasoning that combining the unique morphologies of these two genera would be unlikely to result in the reconstruction of taxon-specific autapomorphies as basal features of the hominin clade (Strait and Grine, 2004). However, both taxa represent highly derived monkeys that may not be suitable as outgroups for hominoids (Benefit and McCrossin, 1997). For example, several Colobus characters, such as the presence of a moderate to large vaginal process of the temporal bone (Strait and Grine; abbreviated henceforth as ‘SG’ Character 36, state 2) and a weak parietomastoid angle (SG 64, state 1), bear superficial (and presumably convergent) resemblance to Homo and Paranthropus that were unlikely to be present in the common ancestor of Pan and Homo. Likewise, Papio displays cranial features that are superficially convergent with derived ingroup morphologies, such as a high degree of basicranial flexion (SG 30, state 3).

Here, we propose that fossil taxa nearer to the base of the hominoid clade represent more appropriate outgroup taxa for defining character polarity in phylogenetic analyses of crown hominoids than do highly derived extant cercopithecoid monkeys. While stem hominids would also be highly informative in this regard, the incomplete nature of their fossil record makes their inclusion here difficult because numerous characters are not preserved, and, therefore, numerous character polarities cannot be assessed using these fossil species. Victoriapithecus macinnesi, a well-known stem cercopithecoid (Benefit, 1993), and Ekembo, a well-known stem hominoid (Hopwood, 1933; Le Gros Clark and Leakey, 1951; Pilbeam, 1969; Andrews, 1985; Kelley, 1986; Andrews and Martin, 1987; Simons and Pilbeam, 2010; McNulty et al., 2015; McNulty and Jansma, 2018), were selected as outgroup taxa for this analysis. Because V. macinnesi is a stem cercopithecoid, it diverged closer to the crown catarrhine node than the extant cercopithecoid outgroups of Mongle et al. (2019). Thus, V. macinnesi had less time to evolve and accumulate many of the derived morphological characters found in extant cercopithecoid taxa, making it more likely to preserve primitive character states and better estimate the ancestral crown catarrhine morphotype. Similarly, as a stem hominoid, Ekembo diverged closer to the crown catarrhine node as well. In addition, Ekembo is phylogenetically closer to the crown hominoid ingroup and likely preserves more primitive character states than extant apes. This allows tree rooting in our analyses with character states that are more informative for crown hominoids than for V. macinnesi alone.

These taxa are also well represented anatomically and therefore allow a polarity assessment for most characters. Victoriapithecus macinnesi is represented by hundreds of craniodental fossils from Middle Miocene deposits on Maboko Island, Kenya. It is the best-preserved stem cercopithecoid (with much of the skull and postcranial skeleton represented), and it has been inferred to have diverged close to the common ancestor of all living monkeys (Benefit, 1999). It is estimated to retain intermediate features between those of modern cercopithecoids and the hypothesized cercopithecoid—hominoid last common ancestor (LCA; Benefit, 1999). Though opinions vary (e.g., Harrison, 2010), Ekembo is most often regarded as a stem hominoid (e.g., Ward, 1997; Alba et al., 2015; McNulty et al., 2015; Hammond and Almécija, 2017; Almécija et al., 2021; Pugh, 2022) and has been supported as such by recent phylogenetic analyses (Nengo et al., 2017; Rossie and Hill, 2018). Given their inferred phylogenetic positions, as well as their representation in the fossil record, both taxa are well suited for use as outgroups in the analysis of hominoid and hominin taxa. We use similar methods in order to directly compare our results with those of Mongle et al. (2019). We hypothesize that adding a stem cercopithecoid (V. macinnesi) and a stem hominoid (Ekembo spp.) as outgroups will improve measures of support for these relationships by reducing the influence of convergent character states from the character matrix.