Although this is the age of the ‘model organism’ there is considerable interest in a much wider diversity of creatures, and techniques from the world of molecular biology are becoming very useful in studying organisms in nature. One area where this has been very successful is in determining mating systems; is a population outcrossing or inbreeding, and, if inbreeding, how much? Determining whether a population is sexual or asexual is even harder, but a recent paper (Signorovitch et al, 2005) has detected signs of sexual reproduction in a species of Placozoa, which are widely accepted as the simplest known free-living animals.

Mating systems have important consequences for populations’ evolution, and the efficacy of natural selection. For instance, the high frequency of homozygotes in highly selfing populations reduces the populations’ effective sizes (Ne), and also means that crossing over only rarely causes genetic recombination (reviewed in Charlesworth and Wright, 2001). Hitch-hiking effects will therefore occur, in which deleterious mutations rise to high frequencies in populations, because the reduced recombination means that some will remain linked to advantageous variants as they spread (although homozygosity ensures that only mutations with small fitness effects will do so). In addition, the removal of deleterious mutations by natural selection further reduces Ne, and makes selection for advantageous variants less effective. If selfing continues long enough, maladaptation may result. Asexual reproduction, without recombination, also leads to accumulation of deleterious mutations through these processes, and to fixation of mutations via the process of Muller's ratchet. However, in such populations, mutations will generally remain heterozygous, so their effects on survival and fertility may be individually mild.

Genetic markers can provide informative data on breeding systems. It is well known that Hardy–Weinberg genotype frequencies imply random mating, that is, absence of inbreeding and of strong differentiation between populations of conspecifics. Selfing and other forms of inbreeding lead to higher homozygote frequencies than expected under random mating, allowing inbreeding to be detected and quantified. Such studies are laborious, requiring genotyping of multiple loci in large samples of individuals collected from nature.

It is thus perhaps not surprising that the breeding systems are still poorly known, even for some important model species. For the plant Arabidopsis thaliana, for instance, selfing rates have been estimated, using allozyme allelic variants, only from populations in Northern Europe. Assuming that all individuals produce the same proportion of seeds by self-fertilisation (the ‘selfing rate’), and that the population has reached its equilibrium inbreeding coefficient under that mating system, genotype frequencies give information on the selfing rate. In these populations (Abbott and Gomes, 1988; Berge et al, 1998), the high selfing rates inferred (about 98–99%) support greenhouse observations that plants readily self-fertilise. However, Northern Europe is a region where selection during the recolonisation after the most recent ice age has probably favoured selfing (Foltz et al, 1982), since densities of animals and plants were probably often low. Populations elsewhere in the world may have higher outcrossing rates.

Caenorhabditis elegans is another important model species with very low outcrossing rates under lab conditions. A recent study has estimated outcrossing rates in several European populations (Barrière and Félix, 2005). Again consistent with lab results, this species is extremely inbreeding, at least in these populations. This study used AFLP markers, a type of DNA-based marker, which has the advantage that large numbers of genetic variants can be scored, in regions scattered around the genome, and thus provide reasonably independent information about what a population is doing. Alleles at different loci were found only in a limited set of combinations, indicating strong linkage disequilibrium, which tells us that recombination has been very infrequent over long periods, from which it was inferred that outcrossing typically occurs only once in many thousands of generations, perhaps the current record for a low outcrossing rate.

It is much harder to determine whether a population is sexual or asexual. As asexual reproduction does not necessarily involve becoming homozygous, an asexual population can show no detectable deviation from Hardy–Weinberg genotype frequencies (though such populations are sometimes recognised because every individual is heterozygous). Over long evolutionary times without recombination, the sequences of the two alleles at each locus may diverge, like alleles isolated in different species (or like alleles at ancestrally homologous loci in the nonrecombining regions of sex chromosomes, see Lahn and Page, 1999). A recent paper (Signorovitch et al, 2005) used DNA sequences to determine that sex probably occurs in a population of a Placozoan – a very difficult study species, being marine and microscopic. cDNA sequences were used to design PCR primers to amplify genomic DNA and sequence the corresponding genes from just a few individuals.

All seven genes included very few sequence variants, showing that each sequenced region is probably a single-copy gene. The alleles have not diverged greatly, so these populations cannot be ancient asexuals. Some individuals were homozygous and some heterozygous, suggesting some sexual reproduction, but excluding a selfing mating system. Some forms of asexuality cannot lead to homozygotes, so in such cases, finding a mix of homozygotes and heterozygotes in a population suggests at least occasional sexual reproduction, but automixis leads to increased homozygosity at some loci (Maynard Smith, 1978). However, no variants were homozygous in all individuals, so, although more loci should probably be studied, this possibility also seems unlikely. Finally, the different loci appear to recombine. If two alleles are present in a population at each of two loci, and if all four combinations of allele are found, this strongly suggests recombination; repeats of the same mutations in the two loci in independent lineages could be an alternative possibility, but seems unlikely in this case because levels of variability are low.

The paper does not analyse the sequence variants (the best analysis if recombination occurs), but haplotypes inferred for the individuals’ genes suggest recombination. If confirmed, the reproduction of these mysterious organisms will be illuminated, even with just these few data showing recombination, although the possibility remains that asexuals arose recently and independently from different genotypes. We can expect the flood of new information on breeding systems of all sorts of organisms to continue, using such approaches (Whitaker et al, 2005). Even rare sexual reproduction, as in many bacteria, is detectable using recombination analyses (eg Maynard Smith, 1999).