Clarifying biotic homogenization.

nouns – ‘hierarchy’, ‘connec-tivity’, ‘evolvability’, ‘complexity’ toname four – are becoming increasinglypopular as evolutionary developmentalbiology tries to find its theoretical feet.With Gerhard Schlosser, an amphibiandevelopmental biologist, GünterWagner has brought together a stellarteam of contributors to discuss theabstract noun du jour, ‘modularity.’In Modularity in Development and Evolution, we canread contributions from 42 authors, many of themdistinguished, about topics as diverse as basic helix–loop–helix transcription factors, data-mining genome-wide expression data, vertebrate limbs, nematode vulvasand Wolbachia symbionts. Unsurprisingly, all authorsagree that modularity is very important to development.The question is: what do they mean?The answer is: many different things. The editors make abrave attempt to keep matters under control by definingmodules as structures or processes composed of tightlyintegrated parts, whilst being relatively autonomous fromtheir surroundings. For developmental geneticists, modulesare thereforegroups ofproteins that worktogether tospecifycells [e.g. the Notch–Delta–Su(H), pathway]. For a quan-titative geneticist, modules can be seen in the distribution ofthe pleiotropic effects of quantitative trait loci, which controlvarious morphological traits. Developmental neurobiolo-gists point to embryologically and functionally distinct unitsof the central nervous system, whereas a nematodegeneticist points to cells. Bioinformaticists see modules in‘synexpression groups’ – clusters of co-ordinately regulatedgenes visible from expression profiling studies; dynamicsystems modelers see them as networks of genes necessaryand sufficient for carrying out a particular function; acomplexity theorist sees them in basins of attraction inrandom boolean networks. And this is only a partial list.The sheer ubiquity of modular things, whatever thosehappen to be, suggests that modules matter to development.Having said that, one has the feeling that many contributorsare cherry-picking the data. They are looking for, andfinding, modules whilst ignoring non-modules. The problemseems to be that the modularity of things is, as the editorspoint out, a matter of degree. However, there is no formaltheory of modularity and, in the absence of that, noconsensus about how to measure it. The result is that weare invariably given the history of a case rather than a senseof its distribution. This is a natural history of modules.What about their evolution? As with more mundaneattributes that organisms might have, say, parental care orwings, evolutionary biologists want to know several thingsabout them. Why do modules exist? Are they the result ofnatural selection? Or can mutation and drift explain theirpresence? If selection, what kind of selection? We also wantknow how they evolve. If a module is found in Caenor-habditis elegans, is it also found in another nematodespecies? A fruit-fly? Humans? Are modules more con-served than non-modules? Has modularity – like complex-ity – increased over the course of evolution?These are important but difficult questions, and thecontributors give a diversity of answers. Taking the last setof questions first, (how do modules evolve), it seems thatsophisticated comparative studies of the evolution ofmodules are some way off. This is not a criticism: suchstudies need lots of data, which hardly exist outside of a fewmodel organisms – but they will surely come. Why domodules exist? Some contributors seem to view modularityasanemergentpropertyofgeneticnetworks.Otherssuggestthat developmental modularity evolves for its variationalproperties – it enables organisms to be more resistant toenvironmentalormutationalperturbations.Thesplithereisanalogous to the Wright–Fisher dispute over dominance.Others again, suggest that modules permit evolvability(i.e. the production of heritable, selectable, phenotypicvariation). This is a clade-selection argument – with allthe weaknesses of such arguments. Many contributorshedge their bets by citing some or all of the above withoutconsidering the matter too deeply. But more carefuldiscussions can be found in at least two papers. Force,Cresko and Pickett argue that ‘genotypic modularity’ – theuse of genes in particular places and times independently ofother genes – can increase simply as a consequence ofCorresponding author: Armand M. Leroi ([email protected]).UpdateTRENDS in Ecology and Evolution Vol.19 No.6 June 2004284 www.sciencedirect.com