Box 1: Why the genetics of adaptation matters for the evolutionary ecology of fatty acids.

Quantitative genetic view of adaptation - Quantitative genetics models generally assume that traits are continuous, normally distributed, and controlled by many genes with small effects. In a model with a single trait, selection moves the trait distribution according to the breeder’s equation: Response to selection (R ) is a product of heritability (h 2) and selection differential (S ) (Lynch and Walsh 1998). Any heritable traits can evolve in response to selection. In cases in which multiple traits are genetically correlated, selection on one trait can bias the evolution of another genetically correlated trait, and the evolutionary trajectories on adaptive landscapes can be biased by a genetic variance-covariance matrix (G-matrix; Lande and Arnold 1983; Schluter 1996; Schluter 2000). Although quantitative genetics models are helpful to predict short-term evolution of highly polygenic traits, genomic studies of fatty acid composition have often identified loci with moderate to large effects (Cesar et al. 2014; Xia et al. 2014; Lemos et al. 2016; Lin et al. 2018; Horn et al. 2020), suggesting that an alternative view of the genetic basis of adaptation may be more useful in some cases.

Genomic view of adaptation - Recent advances in genomic technologies have improved our ability to elucidate the genetic details of adaptation. For instance, studies have helped document how aspects of genetic architecture, such as the number, effect sizes, pleiotropy, linkage, and genomic location of adaptive loci, can influence the speed and reversibility of adaptive evolution (Barton and Keightley 2002). Such studies also aim to identify causative genes. For instance, animal breeding studies have sought to identify genes and quantitative trait loci that control the fatty acid composition of meat (e.g., Kelly et al. 2014; Zhang et al. 2016). However, even when a locus with a major effect is identified, this does not necessarily indicate the presence of a single causative gene. Furthermore, even when a causative gene is identified, it does not necessarily mean that a single mutation causes the alteration of the gene function (Stern and Frankel 2013; Bickel et al. 2011). Therefore, an additional goal of genetic adaptation studies is to identify specific causative mutations (Lee et al., 2014; Remington, 2015). For example, Fads2 duplication in freshwater species derived from marine ancestors allows them to synthesize more n-3 LC-PUFA (Ishikawa et al. 2019). When adaptation occurs via standing genetic variation, it may not be necessary to further dissect it into the levels of individual mutations for predicting how adaptation proceeds. This is because such adaptation occurs by replacement of already-existing alleles with tightly linked adaptive mutations (Barrett and Schluter 2008). However, to understand adaptation by de novomutations, it is essential to determine the number and nature of responsible causative mutations (Stern and Frankel 2013). For example, the bab locus explains over 60% of phenotypic variance of pigmentation in Drosophila , but each single SNP explains only 1% (Bickel et al., 2011). Similarly, different Fads1 and Fads2 gene variants in humans explain between 1-28.5% of the variation in the PUFA content of blood phospholipids (Schaeffer et al. 2006). Once causative mutations that alter fatty acid synthesis are identified, it is then possible to determine whether adaptive evolution has occurred through a few large steps or multiple small steps (Orr 2005).

Box 1 Figure: Relationship between the enzymatic activity and fitness in two different environments (left) and a trajectory of adaptive walk (gray arrow) biased by G-matrix on a two-dimensional adaptive landscape (right). Px, precursor of fatty acid X; Py, precursor of fatty acid Y; Plus (+), the adaptive peak.