Creation science comes as a surprise to many scientists, and thus I suspect that the fact that there is creationist systematics
will come as an even bigger surprise to systematists. Yet creationists do practice a form of systematics, called "baraminology", and for creationist science it is surprisingly rigorous and internally consistent. It employs terminology and methodology not wholly unfamiliar to mainstream systematists (see sidebar, p 21).
The term "baraminology" comes from baramin
, which was constructed from the Hebrew root words bara
(created) and min
(kind) by creationist Frank L Marsh (1941). Baraminology has also been referred to as "discontinuity systematics" (ReMine 1990; Marsh 1941, 1976). Baraminologists consider the baramin to be a taxonomic rank corresponding to the "created kinds" of Genesis. "Intelligent design" creationists are interested in baraminology as a way of quantifying discontinuities in the tree of life (Scherer 1993, 1998; Hartwig-Scherer 1998) and as a boundary between "macroevolution" and "microevolution" (Scherer 1993, 1998), although they tend to shun the term baramin and prefer the term "basic type" (Scherer 1993; Hartwig-Scherer 1998), perhaps because it avoids religious implications. It is also used as a proof of the actions of a designer or special creator (ReMine 1993; Scherer 1998).
The basic idea behind discontinuity systematics is that there are boundaries in the history of life that cannot be crossed. The aim is to find the "discontinuities" in the history of life, or the limits of common ancestry (ReMine 1993). While Marsh may have originated discontinuity systematics in the 1940s, it has been updated and refined to a form that is rapidly becoming one of the most active areas of creationist "scientific" research, and some of its methodology has been applied in near-mainstream research (for example, Scherer 1993). This area of research is also one of the places where "intelligent design" creationist and young-earth creationist "research" overlap.
What is most amazing is the number of traditional systematic methods and terminology that are employed by baraminologists. While they use many of the same methods as most systematists, from cladistics to the Analysis of Pattern (ANOPA) method, they use these tools to identify the "gaps", rather than the connections
in life as most systematists do. This is why baraminologists principally employ phenetic methods of Sokal and Sneath (1963) — which are based on overall similarities in appearance or general features — computing distance matrices for a group of taxa and producing character mismatch
statistics based on the matching coefficient of Sokal and Michener (1958). They see phenetics as useful in determining the biological gaps.
In addition, baraminologists employ cladistics for determining intra-holobaraminic relationships, as well as homoplasy (similarity in form not
attributed to common descent) for separate groups (Robinson and Cavanaugh 1998a). Baraminologists recognize synapomorphy (shared features that are
attributed to common descent) as an example both of a feature that unites a holobaramin, and also of a "discontinuity" among groups. The synapomorphy that diagnoses a group suggests a creative event by God (Wood and others 2003). Baraminologists are very much concerned with having an accurate definition of "kind" because it is vague as commonly used (Awbrey 1981) and because a consistent definition will enable the discovery of the basic created kinds — and ultimately a calculation of the number of animals present on the ark, for young-earth creationists.
Baraminology has had deep roots, but more recently there has been an attempt to codify it into a working method of research for creationist biologists. This culminated in the formation of the Baraminology Study Group (BSG) based at Bryan College in Dayton, Tennessee (http://www.bryancore.org/bsg/index.html
). This group has hosted several conferences on baraminology starting in 1997, and has published a book on baraminological methodology called Understanding the Pattern of Life
(Wood and others 2003). This book offers a concise and relatively complete explanation of baraminology and its practice. Baraminological Taxonomy
The role of discontinuity systematics is to establish the boundaries of common descent. To designate their groups and boundaries, baraminologists employ terminology and designations suspiciously similar to that of most systematists, and in particular to Mayr's evolutionary systematics. The baraminological terminology originally codified by ReMine (1993) and expanded upon by Frair (2000) is shown in the sidebar (see p 21).
For baraminologists, these taxonomic designations have different functions. The ultimate goal is to take polybaramins and break them down into their component monobaramins and their respective holobaramins. These holobaramins could then be placed in apobaramins of structurally similar animals. Baraminologists suggest that it is useful to talk about apobaramins because the holobaramins have many similarities that cross holobaraminic boundaries. Apobaramins are considered useful for studying larger groups of morphologically similar animals. Frair (2000) suggests that humans should be compared with the group most structurally and functionally similar to them: the great apes. There is a sort of cognitive dissonance going on with apobaramins, in that baraminologists are still using the power of phylogenetic inference, even though they deny phylogeny. If the groups are not phylogenetically related, why should baraminologists expect them to be comparable?
Once holobaramins have been determined, the phylogenetic relationships of the in-group members of a baramin could be worked out (Frair 2000). However, baraminologists do not think that displaying these relationships in the forms of trees would be useful. After producing trees, baraminologists have suggested that something other than trees may be more informative for depicting relationships; they suggest that other schemes, such as networks, lattices, or pattern recognition projection plots may do a better job (Frair 2000). Apparently, showing trees is a bit to descent-oriented for comfort. This is probably because representing relationships as descent
, even within a holobaramin might leave people with the "wrong" impression. If people were to think too much about descent within a holobaramin, they may start to think that it can be extrapolated to apobaramins. However, baraminologists propose a slightly different notion of descent within groups:
[D]ifferent members of a holobaramin could have resulted from a sorting out to the offspring of different genes (DNA) from parental organisms. This is a common occurrence today. Or, since the time of creation there could have been some hereditary modifications of the DNA (mutations), and these were passed on to the diverging offspring. Selection in nature could have influenced the potential for survival of the diverse siblings. (Frair 2000)
Finally, it is interesting to note that baraminologists, like phylogenetic taxonomists, claim to eschew "essentialist" thinking, which seems odd given their notions of limited created "kinds"; however, this is because they recognize that diagnostic baraminological features can be lost through variation within a kind. The result is that combinations of other features in the "kind" are used to unite them in a monobaramin (Wood and others 2003). Finding the "Discontinuities" of life
Practitioners of discontinuity systematics claim that it carries no model-based assumptions and therefore can be used independently of creation theories. Their claim is that they do not presuppose discontinuities, but rather follow the data to discover discontinuities where they exist (ReMine 1990). In practice, however, this is not the case. Frequently and explicitly, biblical criteria are used when discontinuities fail to be found among groups that "should be" discontinuous, as in the case of humans and chimps in Robinson and Cavanaugh (1998a). Wood and others (2003) explicitly state that the biblical criteria are paramount and that discontinuities are presupposed because of the separate creation events mentioned in the Bible. Thus, baramins may be defined on a number of criteria, but the "scriptural" criterion takes precedence and is the only one necessary. However, because there are no scriptural definitions for much of life, in those
cases the other criteria are used.
In these cases, baraminologists use a number of membership criteria to determine the boundaries of the holobaramins. These criteria were first proposed as generalizations by ReMine (1990) and have been fleshed out by subsequent works; however, few have been extensively tested or employed as of yet (Wood and Cavanaugh 2003). Save for the biblical criteria, all measures are considered fallible, and thus proponents argue that multiple criteria should be employed when attempting to diagnose a holobaramin (Wood and Cavanaugh 2003; Wood and others 2003). Morphological criteria
Baraminologists have spent perhaps the most research time on the morphological criteria.Within these criteria, baraminologists construct measures of baraminological distance corresponding to character mismatches among the features of the groups they are analyzing. To measure baraminological distance, baraminologists employ a wide range of methods for discovering morphological gaps, borrowing traditional methods such as cladistics and phenetics and developing their own methods. In the computation of such units, traditional cladistic consistency measures and phenetic distance measures are used, along with other multivariate statistical methods. Baraminologists are especially enamored of phenetics (as in Sokal and Sneath 1963) because this approach is particularly amenable to a typological view of life (ReMine 1990). Included in morphological criteria are the identification of morphologies, organs, metabolic pathways, cellular processes, or functions that are unique to a group (sometimes considered synapomorphies) and thus supposedly suggestive of separate origins.
In selecting organisms for morphological analysis to divide monobaramins from apobaramins, baraminologists use typical taxonomic procedures. Baraminologists start with a classification, or according to Robinson and Cavanaugh (1998a), a hypothesized phylogeny (which is ironic given their denial of large-scale phylogenetic relationships), because they assume this group would contain truly holobaraminic groups. Animals of closely allied taxonomic groups are selected as in-groups and nearest neighbor groups are selected as out-groups for biological comparability (Robinson and Cavanaugh 1997).
Baraminologists also use cladistic methods. They compute trees using traditional cladistic software. Groups with high correlations within a bootstrapped dataset are considered potentially holobaraminic and then must be tested for phylogenetic discontinuity of their subgroups. High levels of homoplasy are also considered indicative of separate baramins, which baraminologists propose are a result of separate creations.
Baraminologists consider homologies to exist within
holobaramins; homoplasies are features shared among
holobaramins (Wise 1990). Baraminologists assume that a certain level of homoplasy delineates a phylogenetic discontinuity (Wise 1990), but no discrete criterion has yet been provided. For scriptural reasons, baraminologists think that organisms are too well designed to have true independence of characters (Wood and others 2003), which they argue calls into question the utility of independent character data.
In order to determine the degree of homoplasy, baraminologists compute a "Homoplasy Index" (HI) — the equivalent of 1–CI of traditional cladistic analysis (Kluge and Farris 1969). If the HI is high, then separate baramins are preferred. If the HI is high within a holobaramin, it is proposed to be the result of "gene scattering" from a complex ancestor through hybridization (Robinson and Cavanaugh 1998a; Scherer 1993). As with other baraminological measures, there is no specific measure of what degree of HI represents separate baramins and no explanation of why finding high HIs within
a diagnosed holobaramin is not evidence that it is really an apobaramin. Robinson and Cavanaugh (1998a) note that there is a 0 homoplasy index between humans and apes in their dataset, which they suggests speaks to an imperfect measure by that criteria in some circumstances — namely those in which the analysis does not produce the answer they want. Therefore they caution researchers about using the HI as a criterion. Molecular criteria
Baraminologists have recently become interested in molecular criteria to define a baramin because they offer the chance to search for discontinuities at the "fundamental" level of life (Robinson 1997; but see also Marsh 1971). The reasoning is similar to that of traditional systematists, and baraminology uses some of the same data and analytical techniques. Baraminologists believe that all holobaramins went through a severe bottleneck at the time of the Noachian Flood, so mitochondrial DNA should be an ideal systematic tool (Robinson 1997). Like other systematists, baraminologists download sequences from on-line databases such as GenBank and use typical phylogenetic alignment methods. Baraminologists use taxa that they believe are "phylogenetically distinct" for out-groups in molecular analysis. They then compare sequences by percent sequence difference compared to taxonomic rank, use parsimony distance estimates to construct groupings, and finally evaluate these groupings by bootstrap methods. Molecular methods were pioneered by Robinson (1997) for turtle baraminology. Baraminologists also utilize DNA/DNA hybridization and blood serum reactivity measurements to determine baraminological divisions. Ecological criteria
The ecological criteria were first proposed by Wise (1992), who argued that ecological and trophic differences reflected separate originations or groups. Wise based this on the idea that the taxonomic rank of family reflected created kinds and his observations that the families tend to contain animals with similar ecologies and trophic levels. Thus, different ecological or trophic features should delineate separate baramins. Wood and others (2003) suggest that this criterion would be most useful for single-celled organisms, citing the radically different ecologies and cellular metabolisms found in bacteria and archaea, which, they argue, suggest separate origins. Fossil or stratigraphic criteria
This criterion is a bit hard to understand and employ. For young-earth creationist baraminologists, the stratigraphic record is not reflective of the ancestral history of living things, but rather of deposition during the Noachian Flood. Thus the stratigraphic position of organisms should be irrelevant under this model. Hybridization criteria
Based on the early work of Marsh (1941, 1976), the idea is that the limits of a baramin or basic type can be established for a group of organisms by their ability to hybridize. This is proposed as a testable, definable rank above that of species. It does not matter whether the hybridization is natural, or the offspring is fertile, only that hybridization is possible through some means, including artificial insemination (Scherer 1993). In order to establish these criteria, baraminologists collect and catalog hybrid data, supplemented by some hybridization studies of their own (Scherer and Hilsberg 1982; Scherer 1993). Hybridization potential is correlated with other measures of baraminological distance to test whether groups believed to be monobaraminic are capable of hybridization; if they are, then it provides support for an actual phylogenetic relationship between the organisms (Robinson and Cavanaugh 1998b). One wonders if they are willing to investigate the hybridization criteria for humans and chimps, which was not discussed in Robinson and Cavanaugh (1998a). A hybridization database is available on-line through the website of the BSG (http://www.bryancore.org/hdb/
). Hybridization work is one area where practitioners of "intelligent design" and young-earth creationists overlap. Biblical criteria
The biblical criteria are paramount and trump all other criteria (Wood and others 2003). There are two grades of biblical criteria; the first whether the Bible specifically references the baramin as specially created, and the second whether the Bible implies that it was specially created (Wood and others 2003). There are a number of studies that show how the biblical criteria are employed in baraminological estimates. Robinson (1997) provides a good example for turtles.
First, baraminologists search for any identification of a baramin in biblical texts. In the case of Robinson (1997), it is suggested that turtles are identified in Leviticus. Second, baraminologists determine whether the animal is "clean" or "unclean", thus determining how many pairs were brought onto the Ark. In Robinson (1997), turtles are determined to be unclean; thus only one pair of each turtle baramin would have been required. Robinson also suggests the marine turtles would not have been on the Ark. In the case of humans and primates, Robinson and Cavanaugh (1998a) conclude that even though other criteria cannot separate humans and primates, the biblical criteria specifically states that humans are a separate baramin, so the other data are in effect immaterial.
To aid their quest for discontinuity, baraminologists have developed two semi-original and supposedly objective methodologies, ANOPA and BDIST, which they use along with more traditional systematic methods. ANOPA
One of the membership criteria proposed by ReMine (1990), the "true lineage" can be considered part of the morphological criterion and the stratigraphic criterion. The idea behind this is that organisms could be represented as discrete points in a three-dimensional morphospace. If organisms could be connected by a continuous lineage in morphospace, then they could be considered part of the same baramin. Analysis of Pattern (ANOPA) is a statistical tool developed by Cavanaugh to determine whether such lineages exist (Cavanaugh and Sternberg 2004). Cavanaugh claims that this method is useful for investigating three-dimensional morphological data quantitatively; however, it appears to differ little from principal components analysis with a fancy graphical display, and its measures are suspect (see Dan Bolnick's analysis of ANOPA, p 22
Baraminologists have developed their own analysis software, which performs a distance analysis similar to that of Sokal and Sneath (1963). This is called BDIST (Wood 2001) and is available for free download at http://www.bryancore.org/bsg/bdist.html
. This program is designed to utilize cladistic datasets in NEXUS format as used by PAUP* (Phylogenetic Analysis Using Parsimony; available on-line at http://paup.csit.fsu.edu
BDIST computes the "coefficient of baraminic distance" as originally described by Robinson and Cavanaugh (1998a). This coefficient is a form of the simple matching coefficient of Sokal and Michener (1958). This baraminic distance represents the percentage of characters two taxa share in common. If there is a "chain of positive and significant baraminic distance correlations" connecting all the taxa, then they are monobaraminic (Robinson and Cavanaugh 1998a). Basically, BDIST computes a phenetic distance matrix.
Overall, however, the BDIST methodology has not been extensively applied, and there is no evidence that the algorithmic effects of large datasets, or the role of missing data, have ever been studied by baraminologists. Baraminologists appear to apply an old phenetic method, without really studying how it works. More interestingly the method might not really work at all. In the published applications of the method so far, in no case did it actually distinguish between two baramins. In cases where it returned results baraminologists could live with, they determined a holobaraminic status for the group. This was the case for felids (Robinson and Cavanaugh 1998b), flaveriinae (Wood and Cavanaugh 2001), and fossil and recent equids (Cavanaugh and others 2003). In conditions where it did not return results favorable to baraminologists, other criteria are applied to achieve the desired result. This was the case for humans and primates (Robinson and Cavanaugh 1998a) where BDIST did not show a separation. Instead, the authors employed ad hoc "ecological criteria" to achieve separate baramins, while not discussing the "biblical criteria".
The baraminological groups were originally codified by ReMine (1993) and expanded by Frair (2000).
Holobaramin: All known living and extinct forms understood to share genetic relationships. It is the entire group of organisms related by common ancestry. This would correspond to Mayr's (1963) holophyly or Hennig's (1950) monophyly.
Monobaramin: A group containing only organisms related by common descent, but not necessarily all of them. This could be a group containing one entire holobaramin or a portion of it. This would correspond roughly to Mayr's (1963) monophyly or Hennig's (1950) paraphyly.
Apobaramin: A group consisting of one or more holobaramins. The group of holobaramins may share similar morphology, ecology, and function, but, by definition, not common descent. This may be somewhat like polyphyletic groups.
Polybaramin: A grouping of two or more individuals who are part of at least two holobaramins. It may be a combination of holobaramins, monobaramins, apobaramins, and individuals that by definition do not share a common ancestor. This is consistent with traditional notions of polyphyly.
Baraminologists also recognize a number of taxonomic groupings — archaebaramin (the original created individuals of a holobaramin), neobaramin (the extant individuals in a holobaramin), and paleobaramin (the extinct members of a baramin, or a wholly extinct baramin) — that do not have counterparts in traditional systematics.
Only the study on equids (Cavanaugh 2003) included both fossil and living taxa, and none of the other studies contained datasets with missing data. Therefore, this study served as a template to see how they would investigate fossil and recent morphological datasets. In their treatment of the Evander (1989) data for fossil horses, Cavanaugh and others (2003) removed the missing data by recoding it so that "unknown" data were coded as 0, absence as 1, partial derivation as 2, and presence as 3 and 4 (some characters have more than 2 states). The recoding of unknown data to a specific value that can be used in the analysis makes the dataset use all characters; however, it falsely increases the amount of morphological variation by assigning a numerical value to an unknown. In a sense, the authors artificially create a morphological character state where there is none. This only has minor effects on the overall analysis since it only applies to five characters in a single taxon, so recoding the dataset correctly to include the missing data did not significantly alter the results. However, the effects would be more profound in larger datasets with more widely scattered missing data.
The BDIST software, as configured by Wood, screens out any character with less than "95% relevancy". Why this threshold was chosen is not explained, nor how it is determined. Based on investigation of datasets, relevancy appears to be determined by percentage of missing data for a character relative to total number of taxa. This makes the application of this method to fossil datasets difficult. Surprisingly, characters with no variation (either all 0s or 1s) are not considered "irrelevant" (as they would be in cladistic analyses because they would be "uninformative"). The importance of including characters and taxa with missing data has been shown (Donoghue and others 1989). Recoding missing data as valued, as Cavanaugh and others (2003) have done, however, would have a measurable distorting effect on the results, particularly if the amount of missing data was a larger proportion of the dataset.
When the relative baraminic distances are compared with the phylogenies produced by the datasets, the overall result is a steady, gradual trend in decreasing baraminic distance relative to the phylogeny. These results are comparable to those generated from the Evander (1989) data for fossil horses. Thus, with no significant baraminic distance shifts within the datasets, it could be concluded that dinosaurs and birds belong to the same holobaramin. This makes sense in an evolutionary context: the more transitional features one finds in a set of related organisms, the lower the relative distance between any two taxa will be. In general, including more fossil taxa with transitional morphologies will decrease phylogenic discontinuity, which may explain the datasets that baraminologists have analyzed so far. Conclusion
Despite its use of computer software and flashy statistical graphics, the practice of baraminology amounts to little more than a parroting of scientific investigations into phylogenetics. A critical analysis of the results from the one "objective" software program employed by baraminologists suggests that the method does not actually work. The supremacy of the biblical criteria is explicitly admitted to by Wood and others (2003) in their guidebook to baraminology, so all their claims of "objectivity" notwithstanding, the results will never stray very far from a literal reading of biblical texts. I will give the baraminologists credit in one area: they are up-front about their motives and predispositions and true to their biblical criteria and methodology, which is more than can be said about "intelligent design" proponents.