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

tetrapods ( Ventastega + Ichthyostega + Ymeria + Acanthostega ). However, there are no negative correlations between members of these two groups. Tulerpeton is not positively correlated with any other taxon in the dataset, but is negatively correlated with two members of Group 1 ( Panderichthys , Eusthenopteron ) and one member of Group 2 ( Acanthostega ). Metaxygnathus is neither positively nor negatively correlated with any other taxon. Again, bootstrap values are very low, ranging from 30% to 97% with a median value of 60%. The 3D MDS results (Figure 16) show the same two clusters, with Tulerpeton standing apart from both. The 3D stress was 0.130 with minimal stress of 0.113 at five dimensions. DISCUSSION Garner (2003) claimed that there was a morphological discontinuity between even the most fish-like tetrapods and the most tetrapod- like fishes, and our study provides statistical evidence to support that claim. Leaving aside our analysis of Ahlberg and Clack’s (1998) whole matrix, which includes too many outgroups, three of our analyses show no positive correlations between tetrapods and elpistostegids (Ahlberg and Clack 1998 with 11 taxa; Clack et al. 2016 with 9 taxa; Daeschler et al. 2006) and three show negative correlations (Ruta 2011 with 13 taxa; Swartz 2012 with 10 taxa; Sookias et al. 2014 with 11 taxa). There was only one analysis (Clack et al. 2016 with 13 taxa) in which a single elpistostegid ( Tiktaalik ) was positively correlated with a tetrapod ( Ymeria ), and when this analysis was re-run with fewer taxa and more characters even this positive correlation vanished. Minimally, therefore, our results suggest the presence of two apobaramins: tetrapods and elpistostegids. Our ability to detect discontinuity between the Devonian tetrapods and the elpistostegids is especially noteworthy, given that the Devonian tetrapods possess many fish- like characters and the elpistostegids possess many tetrapod-like characters. Theoretically, taxa that share characteristics of fish and tetrapods could have bridged the gap between these two groups, but our BDC and MDS analyses support separating them into distinct clusters even when such intermediate forms are included. In this respect our results are reminiscent of Wood’s (2010; 2016) finding that statistical baraminology is able to detect discontinuity between humans and non-humans, even though the fossil record includes some humans with ape-like characters and some apes with human-like characters. Some taxa yielded inconsistent results in our study. For example, Elpistostege clustered as expected with other elpistostegids in two analyses (Daeschler 2006; Sookias et al. 2014), but failed to do so in one analysis (Swartz 2012); Elginerpeton clustered as expected with the tetrapods in two analyses (Ahlberg and Clack 1998; Swartz 2012), but failed to do so in one analysis (Sookias et al. 2014); Metaxygnathus clustered as expected with the tetrapods in three analyses (Ahlberg and Clack 1998; Sookias et al. 2014; Clack et al. 2016 with 13 taxa), but failed to do so in one analysis (Clack et al. 2016 with 9 taxa); and Tulerpeton clustered as expected with the other tetrapods in one analysis (Ahlberg and Clack 1998), but in another seemed discontinuous with them (Clack et al. 2016). Moreover, in some of our analyses certain Carboniferous tetrapods clustered with Devonian tetrapods ( Whatcheeria in Ahlberg and Clack 1998, Whatcheeria , Crassigyrinus and Pederpes in Ruta 2011, Pederpes , Diploradus , Whatcheeria and Perittodus in Clack et al. 2016) while in others they clustered separately from them (e.g. Crassigyrinus and Greererpeton in Ahlberg and Clack 1998). Further work will be needed to elucidate the baraminic status of the taxa within these presumably apobaraminic groups. Several possible limitations to the current study suggest themselves. The first concerns the non-holistic nature of some of the datasets, a problem exacerbated by the loss of characters after filtering. For instance, Ahlberg and Clack’s (1998) matrix consisted of mandibular characters only and Ruta’s (2011) matrix of appendicular skeletal characters only. However, missing characters is a perennial problem with fossil data and can only be resolved with the discovery of more fossil material. Moreover, other matrices in our study sampled a greater range of skeletal characters, and in two cases (Daeschler et al. 2006; Swartz 2012) good representation of the character sets was maintained even after filtering. A second concern is the possible non-independence of the character datasets that we analysed. Clearly there is some overlap in the sources used by Daeschler et al. (2006), Swartz (2012), Sookias et al. (2014) and Clack et al. (2016) to construct their matrices, so it is reasonable to ask whether we are actually dealing with six different matrices or merely variants of fewer matrices. However, Garner and Asher ◀ Devonian and Carboniferous tetrapodomorphs ▶ 2018 ICC 467 Figure 15. BDC results for Clack et al.’s (2014) matrix with a subset of 9 taxa, as calculated by BDISTMDS (relevance cutoff 0.75). Closed squares indicate significant, positive BDC; open circles indicate significant, negative BDC. Black symbols indicate bootstrap values >90% in a sample of 100 pseudoreplicates. Grey symbols represent bootstrap values <90%. Figure 16. Three dimensional MDS applied to Clack et al.’s (2016) matrix with a subset of 9 taxa. Devonian tetrapods are shown in blue, elpistostegids in red and other fishes in black.