Original article by Smith et al. 2015, Heredity 1-9.
In summary:
Saltwater tolerance is potentially a case of rapid evolution, occurring multiple times across genera and within species complexes. Often studied through osmoregulation, and via the excretory system. Smith et al 2015 crossed species of mosquito larva that were saltwater-tolerant and freshwater-obligate. Both the BIM and random forest methods supported 6 QTL regions. Maternal inheritance may influence tolerance, as F1 hybrids that were a result of freshwater males crossing with euryhaline females sustained higher levels of saltwater than other combinations For the QTL on the X marker, having only 1 copy rather than 2 seemed to be beneficial. This was seen with a difference in salt tolerance between genders: more males tolerant to salt were hemi/homozygous at the QTL on the X marker, while females tolerant to salt were heterozygous. Ultimately, salt tolerant larva were typically homozygous for the euryhaline genotype at at least 2 of the 6 QTL regions, although it was not consistently the same regions. More than 1 recombination/chromosome event (cross) will be needed for fine-scale mapping. Hybrid sterility will also make fine-scale mapping difficult.
Terms
QTL - loci, or regions in the genome that are involved in the expression of quantitative traits, traits that when partitioned into environmental/genetic origins and variances, have some part that is heritable and therefore coded in the genome. It is a powerful way to find how the genetics underly traits, by associating phenotypes with genomic regions and genotypes. Techniques include screening for candidate genes based on other taxa, SNP-based analyses.
exaptation- when a trait originally evolved no in direct response to selection as is typical of adaptations, but rather via co-expression with other traits that were directly under selection (like saltwater tolerance going along for the ride with drought resistance (Arribas etal 2014)
epistasis (synergistic vs. disruptive): interaction between multiple genes, or one gene with its genetic background (its modifier genes), leading to non-additive effects. Smith et al 2015 tested for it by omitting subsets of data and rerunning analyses to see if QTL results changed.
Segregation distortion: when the segregation ratio of a locus deviates from the expected Mendelian ratio (ie, vary from 1:1 for backcrosses of 1:2:1 for F2s). It is observed because although the markers by definition have no function, they are linked to genes that are subject to gametic/zygotic selection. If the markers actually caused the distortion themselves, they would not be neutral but rather candidate genes for selection. SD influences the viability of individuals with different genotypes of the loci. (Xu et al 2008)
Bayesian interval mapping - Bayesian model selection for mapping multiple, interacting QTLs.
bootstrapping - generally, a resampling method to determine properties of an estimator. Sample the samples.
machine learning inre random forests: learning method, where decision trees are constructed outputs the mode class or mean prediction of the individual trees. Here it was used a a non-parametric technique for classifying samples into fresh/saltwater based on bootstrap sampling of predictors, like SNPs and sex.
telomeres: region of repetitive nucleotides at the ends of chromatin. centromeres also have highly repeietvive DNA sequences and transposable elements. Both are areas that cause issues in inference because of errors, missing data, and abnormal linkage levels.
