Here’s an evolutionary problem: is there any way you can figure whether control mechanisms are global or local by understanding their selection and the selection of their targets? In other words, is there any principled way that we can determine, before the evidence is in, whether a set of phenomena is controlled by one control mechanism or by many, each specific to one or a few targets. This might be similar to the problem in evolutionary psychology about modularity, so I would welcome your thoughts on the matter.
One problem with this question, as I’ve framed it, is that it depends on what you mean by the control mechanism and its target(s). So I’ll give the three contexts in which I have thought about this problem:
There is calibrational change within an individual in response to the environment and then there are also situations in which selection would favour calibrational changes that cross generations. This latter type has been suggested in the context of the thrifty phenotype model and in our discussions about Lamarckism. Besides metabolism many other aspects of an organism’s responses to the environment could undergo this phenomenon (e.g. body form in mollusc ecotypes, immune system responses in humans, etc.).
Are there selectable shortcuts for identifying the relevant sets of these changes and controlling them together? Does nesting affect this? The within-individual set is a subset of the active changes that occur in response to the environment and is less likely to include passive changes (though these might sometimes be adaptive). Similarly the across-generation set is largely a subset of the within-individual set (although critical stages during development could affect this). So a selective shortcut for achieving desired effects, provided the costs of extra-genetic inheritance are not too high, is to include all the members of the relevant superset. (Also the reason the between-generation set is smaller than the within-generation set is that the value of prediction decreases with time, so is there any way to sense this directly too?). Is there a metric? Or would we expect it to be determined in a case by case manner?
2. Competitive signal discrimination and imprinting control
Genomic imprinting, the parent-of-origin specific expression of genes, is thought to result from evolutionary conflicts between maternally and paternally-derived genetic factions operating within one individual. But models have also been developed to look at the genes which apply the imprints or marks (which in turn determine the expression levels of the maternal or paternal alleles of imprinted genes). In certain situations these imprinter genes are selected to imprint, for example, the maternal but not the paternal allele. The paternal allele, though, is selected to be imprinted. Counterfeiting and discriminating mechanisms co-evolve in this situation which is interesting. But again I am curious about control: does this imply that each imprinted genomic region has its own local currency if you like? Or would you expect a specialised imprinter locus that does everything? The evidence so far supports the global model and suggests that one genetic region controls most imprinted genes (Judson et al. (2002) A global disorder of imprinting in the human female germ line. 416(6880):539-42). (See http://cgr.harvard.edu/wilkins_lab/imprinting.html for a great short introduction to the ideas I’ve just touched on).
3. Dosage compensation
Dosage compensation is the process by which the expression (or transcription) of genes on the sex chromosomes is normalised between the sexes and between the sex chromosomes and the autosomes (= all the other chromosomes). In insects, males, which have only one X chromosome, and like us a tiny Y stripped of its genes, up-regulate the expression of genes on their X by a factor of two. This doesn’t happen in females, which have two X chromosomes. So both sexes are the same and the average expression of X-linked genes is the same as the average expression of genes on any given autosome (always present in pairs in both sexes).
In mammals it’s a bit different: females inactivate one of their X chromosomes and this equalises the sexes because both males and females carry just the one functional X chromosome. But very recent evidence (Nguyen and Disteche (2006) Dosage compensation of the active X chromosome in mammals. 38(1):47-53) also shows that mammals, of both sexes, do the same as male insects in that expression from the functional X, in both females and males, is doubled. So in mammals, too, the autosome:sex chromosome ratio is balanced. This is thought to be the result of selection against hemizygous expression (low-level expression caused by having just one copy) of X-linked genes during the early evolution of sex chromosomes (ibid).
Hemizygosity, however, has relatively less impact for some genes compared with others so that it is not clear whether an up-regulation mechanism would evolve for individual genes for which this cost (of hemizygous expression) is significant or whether it would emerge as a general chromosome-wide mechanism. A global mechnanism could pick up distinguishing features of the X chromosome (such features being recognised during X inactivation in female mammals). Just to complicate matters some genes on the X chromsome escape inactivation in female mammals, but the key point is that, as far as I am aware, the evidence is still out for this (we discussed the referenced paper in my department in a recent journal club).
Answering the question
So what’s the deal? Is there a general solution to these sorts of problems or do we have to wait for the experimental work to be done? Is there a way to calculate the costs of global verses local mechanisms or is the modelling best done case by case too? How important is contingency?