Whoa, look at this from the current week’s Nature (7 March 2012), Enni Harjunmaa, et al., “On the difficulty of increasing dental complexity.” Their hypothesis is tremendously significant and its publication represents a genuine breakthrough. The authors argue that known dental mutants (in mammals) invariably show a decrease in complexity, losses of structure, etc. What is never observed, however, are increases in complexity (e.g., novel cusps).
Expressed in street language, they are saying that it’s easy to break something, but hard to make something:
In conclusion, the lack of complex mutants, the high complexity attained adjusting more than one pathway, and the lack of comparable variation in natural populations, all point to a general bias against increase in dental complexity. Yet, the predominant macroevolutionary patterns indicate that ecological factors have been more than adequate in increasing dental complexity. One possible effect of the bias against new cusps could be that, everything else being equal, an increase in dental complexity is slower than a decrease. Of course, yet to be uncovered “complexity genes” may individually increase cusp numbers. Until further evidence, however, we suggest that teeth, and biological structures in general, tend to follow economics of signalling where an increase in complexity beyond the normal variation present in a population requires multiple changes in developmental regulation.
Harjunmaa et al. propose the following hypothesis:
…substantial increase in complexity can also be proposed to require simultaneous changes in several signalling pathways (or multiple changes in a single pathway). This requirement of multiple changes can be conceptualized as “economics of signaling” in which, for example, increasing signalling through one pathway alone may deplete other signals required to produce a greater number of cusps. In genetic terms, this would approximate a polygenic effect.
This is just what Michael Behe has been saying for the past few years. In short, an ID argument — coordinating multiple independent mutations is too unlikely, hence, never observed — in all but name.