Mr. Potato Head and Evolutionary Anomalies

Ewert, W. The dependency graph of life. Biocomplexity 2018(3):1-27. Doi: 10.5048/BIO-C.2018.3.c

Summary. Living organisms are commonly classified in a tree-like pattern based upon a presumed nested hierarchy of traits. The fact that scientists have been able to arrange living organisms into a branching classification has been used as an argument in support of common ancestry. Such a pattern is expected on the assumption of descent with modification, and no other theory is known that predicts a tree-like pattern. Traits that do not fit the pattern of a tree are common, but are regarded as anomalies, producing noise in the data. Anomalies are often explained away as due to gene loss, horizontal gene transfer, convergent evolution, or other causes.

Advocates of intelligent design need a theory that matches the data, including both the tree pattern and the anomalies, more closely than the iconic evolutionary tree. Software development provides some interesting analogies for explaining the pattern of life, which is partly hierarchical and partly non-hierarchical. In creating a new computer program, the programmer has a number of modules available that have been used in other programs. For example, jadom and node-gyp are two JavaScript software modules that depend on a third module, request. These and other modules can be used in a variety of combinations to produce different applications. A program may contain some modules from similar programs, some modules from dissimilar programs, and some unique modules not shared by other programs. These may be visually linked into what is called a dependency graph. A dependency graph differs from a tree because the units may be linked in both hierarchical and non-hierarchical patterns.

Ewert analyzed nine data sets of presence and absence of gene families in a variety of metazoan taxa. Gene families were classified as modules. Some modules were present in many taxa, some in few taxa. The pattern of presence and absences of gene family modules was compared with three potentially explanatory models: a tree, a dependency graph, and a null model of random noise. Next, a hypothetical data set based on strict descent with modification was constructed, and compared with the same three explanatory models. Thirdly, the distributions of modules in a number of JavaScript single page applications were compared with the three explanatory models.

Results showed that the null model did not explain any of the data sets. As expected, the hypothetical data set based on strict common descent fit a tree pattern better than a dependency graph. The JavaScript modules fit a dependency graph better than a tree pattern. Crucially, the gene family modules fit a dependency graph significantly better than a tree. The conclusion is that the best explanation for the pattern of traits in metazoans is the model of common design using modules, represented by a dependency graph. The theory of common descent explains those features that show a tree-like pattern, but not the anomalies. The theory of common design explains both.

Comment. Much effort has been expended in attempts to arrange living organisms in a pattern based on genealogy. Phylogenies produced by these methods are tree-like in structure. However, the tree-like pattern is not as evident as evolutionary theory would predict. There are many exceptions to the expectation of hierarchical nesting of traits. Among prokaryotes, the pattern is so inconsistent with a nested hierarchy that many scientists have abandoned the idea of a grand “tree of life,” at least for prokaryotes.[1] Many scientists still hold that multicellular animals (metazoans) produce a consistent tree-like pattern. However, even among metazoans, inconsistencies are so common that it has been necessary to use a special word - convergence – to refer to them Advances in molecular biology have greatly increased the frequency of detection of evolutionary anomalies, to the point that many evolutionists are looking for a better theory.[2]

In evolutionary theory, organisms that evolved from a common ancestor would be composed only of modules derived or modified from its ancestors. The resulting pattern would unavoidably be a tree. Anomalies would logically falsify the specific tree unless a good explanation can be found.

Yet, despite the repeated apparent falsifications, the presumption of common ancestry is rarely if ever questioned. There is no alternative available to scientists with a materialist philosophy. This paper by Ewert suggests a new approach that may prove more fruitful. We might say, paraphrasing Eldredge and Gould:[3] “anomalies are data.” Ewert’s contribution here is to provide an explanation for the anomalies that does not depend on ad hoc justification, but emerges as a prediction from his model of modular design.

Ewert’s proposal allows the possibility that an intelligent creator could have had a large number of modules available which could be mixed and matched to construct different kinds of animals. Some modules might be used in many species, while other modules might be restricted to a small group of species, or even a single kind. By arranging modules in different combinations, the intelligent creator could make diverse kinds of organisms. It would be possible for a taxonomist to select modules that produce a tree-like distribution pattern in the organisms studied, but there would be a large number of modules that would be distributed in various ways inconsistent with a tree. This model of modular design would predict both the tree-like pattern and the “anomalies” that do not fit the tree. Features that are interpreted as anomalies in the evolutionary tree model are easily explained in the dependency graph model as modules shared with other taxonomic units due to common design.

One might compare the idea of modular design to the children’s toy known as “Mr. Potato Head.” Here, different parts can be mixed and matched to produce a variety of different play faces. Different “potato heads” share modules, and might be arranged in some kind of artificial “evolutionary tree,” but they do not have a history of common descent. Instead, they are produced by common design.

Another metaphor for modular design is the grabbag, filled with multiple interchangeable parts. This was suggested by Stephen Jay Gould, who was noted for his passionate opposition to creationism and intelligent design. Gould compared the evolution of Burgess Shale animals to a process in which a “Great Token Stringer” selected parts from a “grabbag” with a series of compartments, producing a set of animals with mixtures of traits that do not fit the expected hierarchical pattern. Perhaps Gould was more correct than he intended to be. It seems that the grabbag model, much like the potato head model, is more successful than common ancestry in explaining the patterns seen in biodiversity.

L. James Gibson, PhD

Geoscience Research Institute


[1] See, e.g., Doolittle, WF. (2009). The practice of classification and the theory of evolution, and what the demise of Charles Darwin’s tree of life hypothesis means for both of them. Philosophical Transactions of the Royal Society B 364 :2221-2228. Doi:10.1098/rslb.2009.0032

[2] E.g., Pigliucci and M, GB Muller (eds; 2010) Evolution: The Extended Synthesis. Cambridge, MA: MIT Press; Shapiro, JA. (2011). Evolution: A view from the 21st century. Upper Saddle River, NJ: FT Press Science.

[3] Eldredge, N and SJ Gould. (1972) Punctuated equilibria: an alternative to phyletic gradualism. Pp 82-115,in (TJM Schopf, ed) Models in Paleobiology, San Francisco: Freeman Cooper. Their dictum “stasis is data” has become a classic.