Todd C. Wood, Center for Origins Research, Bryan College, 721 Bryan Drive, #7802, Dayton, TN 37321, USA. Tel.: +1 423 775 7277; fax: +1 423 775 7330; e-mail: email@example.com
Senter’s strategy of arguing against creationism using their own methodology focused on demonstrating a morphological continuum between birds and nonavian dinosaurs using classical multidimensional scaling (CMDS), a method used by some creationists to assign species to assist in the detection of phylogenetic ‘discontinuities.’ Because creationists do not typically use CMDS in the manner Senter used it, his results were re-examined using ‘distance correlation,’ a method used to assign species to ‘created kinds.’ Distance correlation using Senter’s set of taxa and characters supports his conclusion of morphological continuity, but other sets of taxa with more characters do not. These results lessen the potential impact that Senter’s strategy might have on creationism; however, it is possible that future fossil discoveries will provide stronger support for morphological continuity between dinosaurs and birds.
Recently, Senter (2010) introduced a novel means of arguing against creationism by using their own methodology to demonstrate conclusions that creationists would otherwise reject. Specifically, Senter used classical multidimensional scaling (CMDS), a common technique in the creationist field of baraminology (i.e. the identification of putative ‘created kinds’), to demonstrate a morphological continuum between coelurosaurian dinosaurs and Mesozoic birds. Senter argued that this continuum supported the grouping of coelurosaurs and birds in the same ‘created kind,’ a conclusion that would be strongly disputed by most creationists (e.g. Woodmorappe, 2003; Silvestru, 2006). Consequently, Senter identified two possible creationist responses to his analysis: rejection of baraminology or acceptance of the dinosaurian origin of birds.
Given ongoing hostility towards the evolution of modern birds from dinosaurs (e.g. Woodmorappe, 2003; Silvestru, 2006), creationists will most likely respond by casting aspersion on or rejecting baraminology. Whether this can be considered an advance in the creation/evolution debate is open to question. Although thoroughly creationist, baraminologists advocate acceptance of widespread speciation even among humans (e.g. Wood, 2010) and have argued for abandoning faulty antievolution arguments (e.g. Cavanaugh et al., 2003). By alienating baraminology from creationism, Senter’s strategy could therefore produce a resurgence of species fixity, which could hardly be called progress.
There is another possible response that Senter does not mention, namely that creationists could dispute his analysis and conclusions. For example, baraminologists use CMDS as a method to visualize morphological distance but not to assign membership in ‘created kinds’ (Wood, 2005a). Consequently, a creationist might argue that Senter’s analysis is incomplete without some additional techniques to show that dinosaurs and birds exhibit significant clustering that could be interpreted as membership in the same ‘created kind.’ This is an inevitable drawback to Senter’s strategy. Ongoing disputes with creationists using their own methodology could be seen as a legitimizing of creationist research, especially when such disputes are published in peer-reviewed science journals.
Here, a baraminological clustering technique, distance correlation, is applied to Senter’s dinosaur/bird data matrix to illustrate the inherent difficulty of following Senter’s approach. If Senter is correct, this method should support the morphological continuum between coelurosaurs and birds. If it does not support his conclusions, the door would be open for creationists to dispute of his work.
Materials and methods
Creationists Robinson & Cavanaugh (1998) introduced the taxon correlation method, wherein a Pearson correlation coefficient is used to estimate the relative similarity of two taxa. Beginning with a set of characters with discrete states, a square distance matrix D with elements dij representing the per cent difference between the character states of taxa i and j is calculated. The relative similarity of taxa i and j is then estimated from D by calculating the Pearson correlation coefficient for rows i and j of D. In theory, two taxa that are similar should have similar distances to the same third taxon and should therefore exhibit positive correlation. Negative correlation should occur between taxa that are very different (i.e. taxa that are close to i will be far from j and vice versa). Statistical significance of the correlation can be estimated using a t distribution with n-2 degrees of freedom, where n is the number of taxa. P-values < 0.05 are considered significant. When two groups of taxa exhibit significant, negative distance correlation, a morphological discontinuity is inferred. It could be argued that distance correlation is a convoluted and unnecessary means of estimating similarity, but it is still the most common creationist method of inferring membership in ‘created kinds.’ Because Senter’s strategy is to use creationist methods to demonstrate evolution, using distance correlation is appropriate whether or not the method is sound, as it is a common creationist method.
Here, distance correlations were calculated for Senter’s 2009 distance matrix, based on 40 characters and 33 taxa. The 2009 matrix contains the full set of taxa known in 2009. Distance correlations were not calculated for any of the taxon sets representing taxa known at previous times. As the most recent taxon set represents the most compelling evidence of morphological continuity between nonavian coelurosaurs and birds, this taxon set was examined exclusively.
Senter used a stringent criterion for inclusion of characters and taxa in his analysis; however, baraminologists emphasize that ‘created kinds’ should be identified using ‘holistic’ data (e.g. Wood et al., 2003). Despite the vagueness of ‘holism,’ it is possible that a larger set of taxa and/or characters could be interpreted as more holistic than Senter’s. Consequently, Senter’s full data matrix was re-examined with less stringent character and taxon inclusion criteria. Taxa with at least 50% of their character states known were included, resulting in a matrix of 42 taxa. Morphological distances were calculated using BDISTMDS at http://www.bryancore.org/bdistmds.html (Wood, 2008). For this larger matrix, BDISTMDS used 187 characters with ≥ 75% of taxa having a known character state to calculate morphological distances.
Another common tactic in baraminology is eliminating outgroups from sets of taxa that are suspected of containing multiple ‘created kinds.’Wood (2005b) argued that distance correlations can be influenced by excessive outgroups that cause spurious positive distance correlations. Alternatively, distant taxa within an otherwise continuous group can show significant, negative distance correlation (Cavanaugh et al., 2003). To mitigate these potential problems, distances and distance correlations were calculated for two subsets of Senter’s complete data matrix. The first subset consisted of fifteen Paraves taxa, and the second subset consisted of eleven taxa of the Avialae and Oviraptorosauria clades. In each subset, only taxa with ≥ 50% of character states known were included, and characters with ≥ 75% of taxa having a known character state were used to calculate distances. BDISTMDS used 157 characters to calculate distances for the Paraves subset and 178 characters for the Avialae + Oviraptorosauria subset.
Distance correlations calculated from Senter’s 2009 distance matrix support his contention that coelurosaurs and Mesozoic birds form a morphological continuum (Fig. 1). Although negative distance correlation can be observed between some taxon pairs, these correlations do not separate any group of taxa from any other. Instead, positive distance correlation can be observed between many taxa that creationists would consider separate creations. For example, nearly all of the Avialae are positively correlated with nearly all of the deinonychosaurs. The composite taxon Protarchaeopteryx + Incisivosaurus is positively correlated with all of the Avialae and deinonychosaurians, two maniraptorans (Falcarius and Ornitholestes) and the tyrannosauroid Guanlong. Likewise, the composite taxon Epidendrosaurus + Epidexipteryx is positively correlated with all three birds, all deinonychosaurians, two maniraptorans (Falcarius and Ornitholestes), two tyrannosauroids (Guanlong and Gorgosaurus) and the compsognathid Huaxiagnathus.
Distance correlations calculated for the larger data matrix of 42 taxa and 187 characters support the existence of morphological continuity between birds and some coelurosaurs (Fig. 2). Positive distance correlation is observed between all birds and deinonychosaurians, but no bird shares positive distance correlation with any nondeinonychosaurian coelurosaur. Nevertheless, the larger matrix supports a much clearer division between taxa than is apparent in Senter’s smaller matrix. Three groups that share positive distance correlation can be identified: the oviraptorosaurians, the Avialae + Deinonychosauria and the remaining taxa (other Coelurosauria + outgroups). Between these groups, only a handful of taxa have positive distance correlation, and many more have significant, negative distance correlation, which would imply phylogenetic discontinuity to a baraminologist.
In contrast to these two full data sets, the two subsets of taxa (Paraves and Avialae + Oviraptorosauria) support a conclusion of discontinuity between birds and other coelurosaurs (Fig. 3). Distance correlation of just Paraves taxa shows significant, negative correlation between the Ornithurae birds and deinonychosaurians (Fig. 3a). Archaeopteryx and Epidendrosaurus are not positively or negatively correlated with any other taxa. The deinonychosaurians are divided into two groups, the Troodontidae + Buitreraptor and Microraptoria + Dromaeosaurinae. Distance correlation of the Avialae + Oviraptorosauria shows significant negative distance correlation between most Avialae and Oviraptorosauria (Fig. 3b). Unlike the distance correlation with just Paraves taxa, all avian taxa share significant, positive distance correlation, and Epidendrosaurus is positively correlated with Archaeopteryx. The significant, negative distance correlation between avian taxa and nonavian dinosaurs would be interpreted by creationists as evidence of discontinuity.
Senter presented the results of his CMDS analysis as definitive evidence that coelurosaurs and Mesozoic birds form a morphological continuum, but the distance correlation results presented here are more ambiguous. Distance correlations calculated for Senter’s set of 33 taxa and 40 characters support the inference of a morphological continuum. With a larger set of taxa and characters, however, three well-marked groups are evident, the Oviraptorosauria, the Avialae + Deinonychosauria and the remaining coelurosaurs and outgroups. The negative distance correlations observed in the two subsets corresponding to the Paraves and the Avialae + Oviraptorosauria would be interpreted by a baraminologist as evidence of a phylogenetic discontinuity between birds and nonavian dinosaurs.
Given these results, it is possible that creationists could simply dispute Senter’s claims. The evidence of discontinuity observed in the subsets could allow a creationist to retain confidence in baraminology while still rejecting the dinosaurian origin of birds. Indeed, it could be argued that baraminology is unsuitable to Senter’s strategy, as it is portrayed as a method of approximating the membership of ‘created kinds’ (e.g. Wood & Murray, 2003). Any baraminology study can be disputed, and none can be considered truly definitive. A study that purportedly shows evidence of major evolutionary transitions using baraminology methods could be legitimately disputed by baraminologists.
Alternatively, the reliance of baraminology on detecting ‘discontinuity’ (i.e. morphological gaps) could in theory render it vulnerable to Senter’s strategy. Even Darwin admitted that morphological gaps exist, but he argued that they could be explained by extinction and the imperfection of the fossil record. Indeed, as Senter (2010) has ably demonstrated, gaps in the fossil record connecting nonavian coelurosaurs and Mesozoic birds that were evident early in the twentieth century have been filled in by more recent fossil discoveries. Even in this study, detected groups represent a wide range of morphology. For example, the group of coelurosaurs apparent in Fig. 2 includes the tyrannosauroids, compsognathids and ornithomimosaurs, thus implying a significant amount of morphological evolution. Perhaps future discoveries will continue to fill in the remaining gaps detected in this study, even to the point where the most convinced creationist must agree that baraminology supports the evolution of birds from dinosaurs.
As argued earlier, however, from a cultural standpoint, the more likely outcome of Senter’s strategy is the rejection of baraminology by creationists. Senter’s strategy of using baraminology to argue against creationism assumes that baraminology and its results are generally accepted by creationists. In fact, baraminology has recently been criticized by some creationists. Although most of these critics still accept the concept of the ‘created kind,’ they object to the ‘distinctly evolutionistic’ nature of baraminological analyses (Menton et al., 2010). For example, Wilson (2010) argued against the results of seemingly noncontroversial turtle baraminology (Wood, 2005b), favouring a greater number of created kinds and thus less evolution. Similarly, a baraminology study purporting to show that the fossil equid radiation represents the evolution of a single created kind (Cavanaugh et al., 2003) has been rejected by Molén (2009, 2010). The recent baraminological analysis of fossil hominids, which placed Australopithecus sediba in the same created kind as modern humans (Wood, 2010), has been disputed by other creationists (e.g. Line, 2010; Menton et al., 2010). The final example is significant, as reactions to the proposal to include A. sediba in the human created kind have called for rejection of baraminology altogether (Menton et al., 2010).
Furthermore, although Senter highlights a number of apparent concessions made by modern creationists as evidence of waning resistance to evolution, creationists remain strongly divided on these subjects. For example, while some creationists accept the geologic column (Snelling et al., 1996), others reject it (Reed & Froede, 2003). Likewise, although some creationists accept beneficial mutations (Bartlett, 2009), others do not (Sanford, 2008). This diversity of opinion among creationists will blunt the impact of any anti-creationist argument that uses creationist methodology, no matter how compelling. Regardless of any research progress (creationist or otherwise), there will always be individual creationists who hold to conservative and traditional antievolution positions.
Thanks to Phil Senter, Michel Laurin, Paul Garner and Dennis Venema for helpful suggestions on earlier versions of this paper.