According to the theory of mate choice based on heterozygosity, mates should choose each other in order to increase the heterozygosity of their offspring. In this study, we tested the 'good genes as heterozygosity' hypothesis of mate choice by documenting the mating patterns of wild Atlantic salmon (Salmo salar) using both major histocompatibility complex (MHC) and microsatellite loci. Specifically, we tested the null hypotheses that mate choice in Atlantic salmon is not dependent on the relatedness between potential partners or on the MHC similarity between mates. Three parameters were assessed: (i) the number of shared alleles between partners (x and y) at the MHC (M(xy)), (ii) the MHC amino-acid genotypic distance between mates' genotypes (AA(xy)), and (iii) genetic relatedness between mates (r(xy)). We found that Atlantic salmon choose their mates in order to increase the heterozygosity of their offspring at the MHC and, more specifically, at the peptide-binding region, presumably in order to provide them with better defence against parasites and pathogens. This was supported by a significant difference between the observed and expected AA(xy) (p = 0.0486). Furthermore, mate choice was not a mechanism of overall inbreeding avoidance as genetic relatedness supported a random mating scheme (p = 0.445). This study provides the first evidence that MHC genes influence mate choice in fish.

We tested the hypothesis that phenotypic parallelism between dwarf and normal whitefish ecotypes (Coregonus clupeaformis, Salmonidae) is accompanied by parallelism in gene transcription. The most striking phenotypic differences between these forms implied energetic metabolism and swimming activity. Therefore, we predicted that genes showing parallel expression should mainly belong to functional groups associated with these phenotypes. Transcriptome profiles were obtained from white muscle by using a 3557 cDNA gene microarray developed for the Atlantic salmon (Salmo salar). A total of 1181 genes expressed in both lake populations hybridized on the array. Significant differential expression between ecotypes was detected for 134 (11.3%) and 195 (16.5%) gene clones in Cliff Lake and Indian Pond, respectively. Fifty-one genes (4.3%) showed parallel differential expression between lakes, among which 35 were expressed in opposite directions. Sixteen genes (1.35%) showed true parallelism of transcription, which mainly belonged to energetic metabolism and regulation of muscle contraction functional groups. Variance in expression was significantly reduced for these genes compared to those not showing directionality in parallelism of expression. Candidate genes associated with parallelism in swimming activity and energetic metabolism based on their level and variance in expression were identified. These results add to the growing evidence that parallel phenotypic evolution also involves parallelism at both the genotypic and regulatory level, which may at least partly be associated with genetic constraints. It also provides further evidence for the determinant role of divergent natural selection in driving phenotypic divergence, and perhaps reproductive isolation, in the adaptive radiation of lake whitefish. This study adds to a nascent field employing microarrays as powerful tools for investigating the evolutionary processes of adaptive divergence among natural populations.

aflpop is a population allocation and simulator program based on amplified fragment length polymorphism markers. The allocation method is an adaptation of Paetkau's method for co‐dominant alleles. Besides population allocation of specimens of unknown origin, re‐allocation of sample genotypes, as well as allocation of artificial (Monte Carlo) specimens, may be run to estimate expected rates of correct allocations. Thanks to its embodied simulator, aflpop can provide information on the rates and types of incorrect allocations and on empirical distributions of likelihood statistics. A filtering procedure within aflpop allows the selection of loci according to user‐defined criteria.

Individual-based population assignment tests have thus far mainly relied on the use of microsatellite loci. However, the logistic difficulty of screening large numbers of loci required to reach sufficient statistical power hampers the usefulness of microsatellites in situations of weak population structuring. Amplified fragment length polymorphisms (AFLP) represents an alternative for overcoming this logistical issue as the technique allows the user to characterize a much larger number of loci with a comparable analytical effort. In this study, an assignment test based on maximum likelihood for dominant markers was used to investigate the potential usefulness of AFLP for population assignment. We also compared assignment success achieved with AFLP with that obtained using microsatellites in a case study of low population differentiation involving whitefish (Coregonus clupeaformis) sympatric ecotypes. The analytical investigation showed that the minimum number of AFLP loci required to reach an assignment success of 95% stood within values that are easily achievable in many situations. This also showed how assignment success varied according to the number of AFLP loci used, their absolute frequency and their frequency differential and sampling errors, as well as the number of putative source populations. The case study showed that given a comparable analytical effort in the laboratory, AFLP were much more efficient than the microsatellite loci in discriminating the source of an individual among putative populations. AFLP resulted in higher assignment success at all levels of stringency and the log-likelihood differences between populations obtained with AFLP for each individual were much larger than those obtained with microsatellites. These results indicate that research involving individual-based population assignment methods should benefit importantly from the use of AFLP markers, especially in systems characterized by weak population structuring.

No abstract