The lack of robustness of models of the maintenance of polymorphism in a heterogeneous environment which has been pointed out by Maynard Smith & Hoekstra (1980), applies also to models based on habitat selection, on temporal variation and on density-regulated selection. Only if (partial) dominance ‘switches’ between environments such that the fitness of the heterozygote is always close to the favoured homozygote, is there reasonable robustness. This is true for all models considered. It is argued that there are good reasons for supposing that the favourable allele at a locus may show dominance, although the experimental evidence is still scanty.

The combined effect of multi-niche selection and cyclical selection is studied in a deterministic model. Sufficient conditions for protected polymorphism are derived for diploid populations with and without dominance and for haploid populations. These conditions appear to be broader than in the cases of only multi-niche selection or only cyclical selection.

A deterministic model of cyclical selection in randomly mating populations is studied. Sufficient conditions for a protected polymorphism, which are for the special case of alternating selection also necessary conditions, are obtained using a simple graphical approach. The most important condition requires ‘marginal overdominance’ (Wallace, 1968); the other conditions seem hard to satisfy in a natural situation. Furthermore it is shown that the cyclical selection model can be regarded as a special case of a frequency-dependent selection model (Cockerham et al. 1972). Using this property, a mean fitness function for the cyclical selection model is derived. Generally, the mean fitness will not be maximized under cyclical selection. The relevance of the model to the problem of the role of cyclical selection in the maintenance of genetic polymorphism in natural populations is discussed. It is concluded that this relevance is probably rather limited with regard to the creation of protected polymorphism, but that the influence of cyclical selection on transient polymorphisms might be more significant. An approximate formula for the time needed for a given change in gene frequency under cyclical selection is derived. It appears that cyclical selection can extend considerably the time during which a transient polymorphism persists, especially if the selective differences in the different environments are of the same order of magnitude and of opposite sign.

A set of aspergillus strains from culture collections and wild-type black aspergilli isolated on non-selective media were used to validate the use of media with 20% tannic acid for exclusive and complete selection of the black aspergilli. The 20% tannic acid medium proved useful for both quantitative and qualitative selection of all different black aspergilli, including all recognized species: A. carbonarius, A. japonicus, A. aculeatus, A foetidus, A. heteromorphus, A. niger, A. tubingensis and A. brasiliensis haplotypes. Even higher concentrations of tannic acid can be utilized by the black aspergilli suggesting a very efficient tannic acid-degrading system. Colour mutants show that the characteristic ability to grow on high tannic acid concentrations is not causally linked to the other typical feature of these aspergilli, i.e. the formation of brown-black pigments. Sequence analysis of the A. niger genome using the A. oryzae tannase gene yielded eleven tannase-like genes, far more than in related species. Therefore, a unique ecological niche in the degradation of tannic acid and connected nitrogen release seems to be reserved for these black-spored cosmopolitans.

The ectomycorrhizal fungus Hebeloma velutipes consists of two biological species (BSP 16 and 17). Within BSP 17 a dikaryon was found with two divergent types of the ribosomal Internal Transcribed Spacer (ITS1 and 2). The two ITS types segregated in monokaryotic progeny of that dikaryon, showing that these different ITS types represent different alleles at homologous rDNA loci in the two nuclei. RFLP analysis of a number of strains of BSP 17 showed that the polymorphism is widespread in Europe. There was no deficiency of the heterokaryotic type, demonstrating that ITS divergence in this species is not correlated with reduced intercompatibility. A strain from North America, not assigned to a biological species, showed the same polymorphism. Cladistic analysis of the two ITS sequences showed that they were not sister groups. One of the ITS types formed a monophyletic group together with the ITS type of BSP 16, the other type formed a clade with the ITS type of H. incarnatulum (BSP 18). BSP 16 and 17 showed partial intercompatibility. However, several lines of evidence suggest that the polymorphism of BSP 17 is not the result of frequent and continuing hybridisation with BSP 16. Instead, we give arguments for the hypothesis that the polymorphism evolved in allopatry and that the two types have come together relatively recently. The results of the polymorphism indicate a potential problem for molecular identification of fungal species based on ITS fingerprinting. The results also show that no generalisations are possible about the relation of speciation (the formation of BSP) and nuclear ITS divergence.

Information on the nature of epistasis between alleles affecting fitness is hardly available, but relevant for, among other issues, our understanding of the evolution of sex and recombination. Evidence of synergistic epistasis between deleterious mutations is support for the Mutational Deterministic hypothesis of the evolution of sex, while finding antagonistic epistasis between beneficial alleles would support the Environmental Deterministic hypotheses. Both types of epistasis are expected to cause negatively skewed fitness distributions of full-sib offspring from a sexual cross. Here, we have studied the form of the distribution of a variety of quantitative characters related to fitness by searching the literature. The fitness traits encountered include the mycelial growth rate in fungi, and earliness, resistance against pathogens, seed number, and pollen fitness in plants. Fitness-related traits in plants show almost exclusively negative skewness, while the results for fungal species are more ambiguous. Possible sources of negative skewness other than epistasis, such as recessiveness of deleterious alleles or a negatively skewed error variance, were tested and found to be unimportant. We argue that these results suggest the existence of synergistic epistasis between deleterious alleles or antagonistic epistasis between beneficial alleles in plants, which is general support for the currently popular hypotheses of sex and recombination, but does not distinguish between them.

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.