Most evolutionary geneticists would agree that the major problems of the field have been solved. We understand both the nature of the mutational processes that generate novel genetic variants and the populational processes which cause them to change in frequency over time — most importantly, natural selection and random genetic drift, respectively. While there is plenty of debate about the relative importance of different processes in patterns of evolutionary change, no serious evolutionist will now defend onceprevalent views such as orthogenesis (predetermined evolution) or the inheritance of acquired characteristics. Indeed, jaded researchers might claim that all the soluble and interesting problems have been disposed of, leaving us with the insoluble and the trivial, as Wolpert suggests for developmental biology [1]. My own feeling is that there are still many intellectually exciting and potentially soluble problems, although we will never again come up with concepts as fundamental as those formulated by the ‘founding fathers’ of population genetics (Fisher, Wright and Haldane), or do experiments as path-breaking as Dobzhansky’s demonstration of natural selection acting on polymorphic chromosome inversions. The two most pressing questions I would put to a good fairy godmother of evolutionary genetics are these. First, what is the frequency distribution of selective effects among new mutations — how often do they have positive, negative or neutral selective effects, and how large are these effects? And second, what is the total rate per genome at which mutations that have a sizeable impact on fitness appear? I would like my questions answered for my favourite organism, Drosophila melanogaster, as much of the empirical work relevant to these questions has already been done using this delightful creature. Answering the first question would bring us close to understanding the role of selection in controlling the frequency within a population of variants at particular nucleotide sites. This issue was first raised thirty years ago, with the discovery of large amounts of within-population variation at the amino-acid sequence level [2], but has never been conclusively answered. The discovery of large amounts of silent and noncoding site variation has simply modified the question: are most nucleotide changes so slightly advantageous or disadvantageous that their fate is controlled mainly by genetic drift? Or is a significant fraction of variants subject to selection that is so weak as to be undetectable by direct measurement but strong enough to exert an influence on their evolutionary fate? We simply do not know, despite increasing statistical evidence from case studies of sequence variation and evolution of both DNA and protein which suggest a role for selection [3]. The other reason for asking this question is that many features of genetic systems are likely to have been shaped by the input of deleterious mutations at loci scattered across the genome [4]. From several Drosophila experiments, along with indirect evidence from other sources, it seems likely that a newly formed zygote in a higher eukaryote has a probability close to one of carrying a new deleterious mutation with an effect on its fitness of the order of one or two per cent relative to the fitness of nonmutant individuals [5]. If true, this has major implications for the evolution of sexual reproduction and genetic recombination, diploidy versus haploidy, outbreeding versus inbreeding, sex chromosome evolution and the evolution of ageing [4]. In addition, rates of evolution and variation for less strongly selected mutations may be influenced by strongly deleterious alleles at neighbouring sites on the chromosome; we may have to modify current theories that treat loci as essentially independent units before we can properly test hypotheses about the causes of nucleotide site evolution and variation [6]. In the absence of good fairy godmothers, I am much more sanguine about answering the second question than the first; it essentially requires brute-force application of mutation-accumulation methods, on a sufficiently large scale that one can have confidence in the resulting estimates. The first is more elusive, but will no doubt provide population geneticists with harmless employment for years to come.
Read full abstract