Viruses are common in asexual Aspergilli but not in sexual Aspergilli. We found no viruses in 112 isolates of the sexual Aspergillus nidulans. We have investigated factors that could play a role in preventing the spread of mycoviruses through populations of A. nidulans. Experiments were performed with A. nidulans strains infected with viruses originating from A. niger. Horizontal virus transmission was restricted but not prevented by somatic incompatibility. Viruses were transmitted vertically via conidiospores but not via ascospores. Competition experiments revealed no effect of virus infection on host fitness. Outcrossing was found to limit the spread of viruses significantly more than selfing. It is concluded that the exclusion of viruses from sexual Aspergilli could be due to the formation of new somatic incompatibility groups by sexual recombination.

The rolA gene of Agrobacterium rhizogenes contains in its untranslated leader region a spliceosomal intron, which is spliced in Arabidopsis and in Nicotiana tabacum. Expression under the control of the 35S promoter from cauliflower mosaic virus of a rolA gene derivative defective in splicing still causes alterations of growth in transgenic tobacco plants. Splicing of rolA mRNA is required for efficient expression of the rolA phenotype in vivo. Moreover, splicing is required for efficient in vitro translation of the rolA mRNA. In contrast, expression of a 35S-rolA gene derivative with the ATG initiation codon replaced by ATA does not cause any phenotypical alteration. Mutations leading to amino acid substitutions at positions 37 and 40 of the rolA coding region were isolated as null mutants in Arabidopsis plants transgenic for the rolA gene. However, when expressed in tobacco under the control of the 35S promoter, they cause a rolA phenotype reduced in the expressivity of its traits. The molecular characterization of rolA mutants might be useful for understanding the biochemical function of the rolA protein.

In the evolutionary process during which Drosophila sechellia became specialized on a toxic fruit (morinda), a spectacular decrease in female reproductive capacity took place when compared with the species' generalist relatives D. mauritiana and D. simulans. Comparisons of species and interspecific crosses showed that two different traits were modified: number of ovarioles and rate of egg production. During the conservation of a D. sechellia strain on usual food, adaptation to laboratory conditions led to an increase in the rate of oogenesis but not in ovariole number. Comparison of F1 and backcross progeny also suggests that the two traits are determined by different genes (ovariole number has already been shown to be polygenic). When morinda is available as a resource, the low rate of egg production in D. sechellia is partly compensated by a stimulating effect, while an inhibition occurs in D. simulans. It is assumed that D. sechellia progressively adapted itself from rotten, non-toxic morinda to a fresher and more toxic resource. During this process the rate of oogenesis evolved from an inhibition to a stimulation by morinda. Simultaneously a spectacular decrease in ovariole number took place, either as a consequence of stochastic events related to the small population size of D. sechellia and a metapopulation dynamics, or as an adaptive process favouring dispersal capacities of the female.

Koala (Phascolarctos cinereus) populations in southern Australia have a history of bottlenecks – earlier this century the species became extinct in South Australia, and almost so in Victoria. Subsequently large numbers of animals from island populations (founded from very few animals) have been translocated back to mainland sites and to other islands in the region. As part of a larger study of the genetic structure of koala populations in southern Australia, we have undertaken a survey of mitochondrial DNA restriction fragment length polymorphism (mtDNA-RFLP) variability. Genomic DNA from 91 koalas from five populations was examined using 23 restriction enzymes, and mtDNA fragments were detected using a domestic cat full-length mtDNA clone. Only one of the enzymes, TaqI, revealed polymorphism – a relatively low amount of variation compared with other mammals, although low mtDNA-RFLP variation has also been reported in Queensland koalas. French Island and populations established predominantly from French Island immigrant koalas, either directly or via other island populations, were indistinguishable by haplotype frequencies. The mtDNA data are thus consistent with the interpretation that the koala translocation programme has homogenized gene frequencies amongst those populations involved. South Gippsland is not recorded as having received translocated koalas directly, and has significantly different mtDNA-RFLP haplotype frequencies from all other populations examined. The fact that this distinction was not previously observed in nuclear gene frequencies may reflect predominantly male-mediated dispersal in koalas.

A modification of the neighbour joining method of Saitou & Nei (1987) is shown to be applicable to the ordering of genetic markers. This neighbour mapping method is compared with some other procedures for ordering genetic markers using both real and test data sets. The limitations and likely errors associated with the use of neighbour mapping are discussed. The speed and simplicity of this method commend its application, as does its concurrence with other mapping methods.

A new estimator is proposed for the parameter C=4Nc, where N is the population size and c is the recombination rate in a finite population model without selection. The estimator is an improved version of Hudson's (1987) estimator, which takes advantage of some recent theoretical developments. The improvement is slight, but the smaller bias and standard error of the new estimator support its use. The variance of the average number of pairwise differences is also derived, and is important in the formulation of the new estimator.

A simple model of migration between two populations, each in a balance between mutation and stabilizing selection on a polygenic trait, is explored. Below a critical migration rate, genetic differences between the two populations can be maintained, even if the populations are selected towards the same phenotypic optimum. Gene flow then maintains genetic variance within each population. For this process to account for heritable variation, there must be some mechanism that causes divergence. The possibility that fluctuating selection could lead to the initial differentiation of the populations is explored.

We used Drosophila melanogaster to test for compensatory control of cell area and cell number in the regulation of total wing area. In two random bred wild-type base stocks collected from different geographic locations we found a negative association between the area and the number of cells in the wing blade. Three replicate lines were selected for increased or decreased wing area, with cell area maintained at the same level as in the three controls. After eight generations of selection, despite a large and highly significant difference in wing area between the large, control and small selection lines, cell area did not differ significantly between them. Rather, the difference in wing area between selection regimes was attributable to differences in cell number. Over the course of selection, the initially significant negative correlation between cell area and cell number in the wing increased, providing evidence for compensatory regulation of cell area and cell number. As a result of the increasingly negative association between the two traits, the variance in wing area declined as selection proceeded. It will be important to discover the mechanisms underlying the compensatory regulation of cell area and cell number.

We compare the powers of three methods for the QTL analysis of non-normally distributed traits. We describe the nonparametric and the logistic regression approaches recently proposed in the literature and study the properties of the standard regression interval mapping method when the trait is not normally distributed. It is shown that the standard approach is robust against non-normality and behaves quite well for both continuous and discrete characters. The loss of power compared with the nonparametric or the logistic approach is generally minor. Moreover, the least squares estimation procedure of the regression interval mapping is not affected by departure from normality. The use of other approaches could be restricted to extreme cases where the trait distribution is very skewed.