Nobody here is disputing that adaptive loss of function mutations occur, nor is anyone here disputing that they can predominate under many circumstances, including in the wild. What is being disputed is Behe’s overall thesis, which is not only that this process occurs, but that it is the net, long-term outcome of the evolutionary process basically under any circumstance, such that the complexity we see in life couldn’t be produced by evolution as we currently understand it. There’s just no good evidence for such a grandiose claim. There is lots of evidence against it. He first derived this “net” effect view of evolution largely from a review of a handful of studies of evolution in a few simple, synthetic lab environments where organisms were taken from their enormously complex wild-type environments and evolved in simple, constant, flask environments.
Evolution is a combination of multiple processes with their own varying biases that can fluctuate in magnitude over time, as conditions change. Some times conditions change and become simpler, with fewer or no ecological opportunities that reward new functions, and then you just get a sort of optimization by any unnecessary genetic material and gene expression being shut off and eventually suffers deactivating if not deletion mutations. This can of course not go on indefinitely, as eventually no more functions or genes can be lost without incurring strong fitness penalties, as you’ve essentially removed everything non-essential and which doesn’t positively contribute to fitness in that environment.
Once this floor is reached, there is no other way to go but either to stay there, or to increase in complexity again when conditions allow (and you might ask entirely legitimately, what conditions do allow this?).
Of course, the “cost” of excess genetic material depends largely on population size. Generally speaking, selection for single-celled organisms such as bacteria is much more efficient at removing excess DNA, because of their huge population sizes. This situation is often times much more relaxed for large multicellular eukaryotes. Hence for multicellular eukaryotes, more complexity is allowed.
The “streamlining” effect of selection can also be counterbalanced by other processes that can have inherent tendencies for complexification. There are now multiple well-known examples of constructive neutral evolution driving up complexity. Microsatellite DNA amplifications, actively transposing retrotransposons, and other types of selfish genetic elements can drive up genome size and add to the constructive neutral evolution process.
Then there are circumstances where the environment starts selecting for novel functions. When you’re down near that floor of complexity where nothing else can be lost without incurring a fitness cost, new ecological opportunities can open up and reward innovations(gain of function mutations). A novel carbon compound enters the environment, and then suddenly this enzyme that has this weird inherent side-activity that never did anything is suddenly useful, so now amplifications of this enzyme gene are beneficial, and now there’s selection for enhancing this side-activity in some of these duplicates and so you get complexification by gain-of-function mutations through the innovation-amplification-divergence process. You now have two enzymes each specialized towards one function each, where before you just had one. Add more novel environmental challenges and rewards, and you get the opportunity for selection to reward more innovation.
)
