Incremental Indispensability Revisited
A common rebuttal to the argument for intelligent design from irreducible complexity is the hypothesis of incremental indispensability (e.g. Draper, 2002). The phrase was coined, I believe, by William Dembski in his book No Free Lunch in response to evolutionary biologist H. Allen Orr's critique of Michael Behe's Darwin's Black Box.
The basic idea is that although components of a system may be essential in the present, they may have been dispensable in the past. Perhaps the addition of components was at first merely beneficial, but later became essential due to the modification, or even loss, of other components. It is envisioned, therefore, that irreducibly complex systems could have arisen by a direct Darwinian pathway after all, by adding components that are at first only adaptive but are later rendered essential.
How convincing is this hypothesis? Does it have any empirical basis? And are there any real-world examples of such a process in action? Actually, there are. Let me offer two illustrations.
Tethering the Z Ring
Our first example pertains to bacterial cell division (discussed in more detail here) and the tethering of the FtsZ proteins (that assemble at the division-septum) to the inner membrane by the proteins ZipA, FtsA and ZapA (Huang et al., 2013; Hale et al., 2011 Pichoff and Lutkenhaus, 2002). Among the gamma-proteobacteria (which include Escherichia coli and Salmonella enterica), both ZipA and FtsA are essential, while ZapA contributes to the efficiency of the process but is not essential (Hale et al., 2011). It is interesting to note that ZipA is not found outside the gamma-proteobacteria, and it appears that "the requirement for ZipA can be bypassed completely by a single alteration in a conserved residue of FtsA" (Geissler et al., 2003). There is also evidence to suggest that elevated concentrations of FtsA can compensate for a lack of ZipA. For example, FtsA is present in significantly higher concentrations in the Firmicute bacterium Bacillus subtilis than in Escherichia coli (Feucht et al., 2001).
Could this be a convincing counter to the argument from irreducible complexity? Could it be that ZipA was at first useful in the Gamma-proteobacteria, but was later rendered essential by a slight modification, or decrease in concentration, of FtsA, thereby leaving a false impression regarding the irreducibly complex core of the bacterial division apparatus?
This example, illustrative of all such cases with which I am familiar, fails to convince. Why? Because although a particular anchor protein (such as ZipA) is not necessarily required for successful cell division, at least some anchor protein is certainly required. In other words, while an anchor is required for the system (indeed, is essential), the precise identity of the protein fulfilling that role is not important -- it could even change. Thus, a strong case could still be made that this is an example of an irreducibly complex system. Unless FtsZ and a tether are present together, the system is useless.
The Case of Syncytin
A second example is the fusogenic syncytin proteins that fuse placental cells together to form the syncytiotrophoblast and which are known to be essential for placental formation. These proteins are coded for by the envelope (env) gene of an endogenous retroviral insert. These proteins are absolutely critical for placental development in humans and mice. The different kinds of Syncytin protein are called "syncytin-A" and "syncytin-B" (found in mice); "syncytin-1? and "syncytin-2" (found in humans). Interestingly, although serving exactly the same function, syncytin-A and syncytin-B are not related to syncytin-1 and syncytin-2. Syncytin protein plays the same function in rabbits (syncytin-ory1). But rabbit syncytin is not related to either the mouse or human form. These ERVs are not even on the same chromosome. Syncytin-1 is on chromosome 7; syncytin-2 is on chromosome 6; syncytin-A is on chromosome 5; and syncytin-B is on chromosome 14. For some reviews of this topic, see Dunlap et al. (2006); Dupressoir et al. (2005); and Harris (1998).
Now, target-site duplication is the hallmark of insertion by integrase (as a consequence of the repair by the host cell's DNA repair proteins of single-strand gaps at each end of the inserted sequence formed upon element insertion). We can use this to infer that these elements are indeed inserts: They are not intrinsic to the genome. So, in this case, an ERV has acquired function subsequent to its insertion, and has apparently been later rendered essential.
Could this serve as a counter to irreducible complexity? Here is the question to ask: Did the system function before the insertion of the retrovirus? If the molecular machinery in which the syncytin proteins are functioning is determined to be irreducibly complex, then the theory of intelligent design would predict that it did not function -- unless of course some other protein was fulfilling the role and was subsequently replaced by the syncytin, as per the previous scenario regarding ZipA and FtsA.
While an intuitively appealing rebuttal to the notion of irreducible complexity, the thesis of incremental indispensability lacks merit. To illustrate the conceptual difficulty with this rejoinder, consider a rattrap with two springs (rather than one spring like a mouse trap) to give it extra force for catching bigger rats. The second spring is certainly helpful, but not essential. Nonetheless, it is conceivable that modification or even loss of one spring could render the other one essential for the system's functioning. The precise identity of the trap's spring, therefore, is not important -- it can even change. But the trap still requires at least one spring. This seems to me to be the uniform problem with all examples of such a process that I am familiar with. But I am of course happy to be pointed to a better example, with which I have yet to become acquainted.