Recombinant inbred line panels inform the genetic architecture and interactions of adaptive traits in Drosophila melanogaster

Abstract

The distribution of allelic effects on traits, along with their gene-by-gene and gene-by-environment interactions, contributes to the phenotypes available for selection and the trajectories of adaptive variants. Nonetheless, uncertainty persists regarding the effect sizes underlying adaptations and the importance of genetic interactions. Herein, we aimed to investigate the genetic architecture and the epistatic and environmental interactions involving loci that contribute to multiple adaptive traits using two new panels of Drosophila melanogaster recombinant inbred lines (RILs). To better fit our data, we re-implemented functions from R/qtl (Broman et al. 2003) using additive genetic models. We found 14 quantitative trait loci (QTL) underlying melanism, wing size, song pattern, and ethanol resistance. By combining our mapping results with population genetic statistics, we identified potential new genes related to these traits. None of the detected QTLs showed clear evidence of epistasis, and our power analysis indicated that we should have seen at least one significant interaction if sign epistasis or strong positive epistasis played a pervasive role in trait evolution. In contrast, we did find roles for gene-by-environment interactions involving pigmentation traits. Overall, our data suggest that the genetic architecture of adaptive traits often involves alleles of detectable effect, that strong epistasis does not always play a role in adaptation, and that environmental interactions can modulate the effect size of adaptive alleles.

Keywords: Drosophila melanogaster; Quantitative Trait Loci (QTL); adaptation; epistasis; quantitative genetics.

Figure 1.

The number of QTLs for each trait ranged from 0 to 5. In this figure, only the traits with at least one identified QTL are shown (phenotype distributions for each genotype at each QTL peak window are shown in Figures S3, S4); the traits without significant QTLs (Table 1) can be seen in Figure S5. Each panel shows the −log of the (trait-specific) p-value for the LOD score of the genomic windows. A4 Background (Figure 1E) is shown here with no QTL above the threshold, but one marginally significant QTL (p-value = 0.053, Table 2). Windows filtered out for ancestry skew are given a value of −0.25. The red dashed line represents the 0.05 p-value cutoff based on 10,000 permutations. The color of the dots represents the chromosome arm of each genomic window. Note that for wing length, a single QTL spans a broad low recombination centromeric region between 2L and 2R.

Figure 2.

Simulation analysis indicating that strong positive epistasis or sign epistasis should have been detectable for the empirical data. Simulations based on randomly permuted individual genotypes and simulated epistatic effects (see Materials and Methods) were conducted to assess our ability to detect varying models of epistasis based on either the lowest empirical epistasis p-value (red series) or the Fisher-combined epistasis p-value across 13 analyzed QTLs (blue series). Here, the epistasis factor (x-axis) represents the multiplier that a modifier locus exerts on the primary QTL’s effect. Thus, negative values represent sign epistasis (pink shaded area), zero represents masking epistasis, values between 0 and 1 indicate negative epistasis (yellow shaded area), and values above 1 indicate positive epistasis (green shaded area). The y-axis shows the proportion of occurrences out of the 1,000 resampled instances in which a given model of epistasis yielded a lowest p-value lower than the observed 0.183 (red), or else how often the combined Fisher p-value was lower than the observed 0.763 (blue).

Figure 3.

The reaction norms of the four non-overlapping pigmentation QTL show darker phenotypes for flies raised at 15 °C than at 25 °C and for flies with Ethiopia ancestry alleles, corroborating expectations. (A) Q1: A4 Background at 25 °C, with peak centered near 21.6 Mb on 3R. (B) Q2: Mesopleuron at 15 °C, near 16.8 Mb on 3R. (C) Q3: Mesopleuron at 25 °C, near 17.6 Mb on 3R. (D) Q5: Stripe ratio at 15 °C, near 1.3 Mb on the X. The y-axis shows the mean phenotype for each genotype and temperature treatment; a higher number is a darker phenotype. ZZ = Zambia homozygous, EZ = heterozygous, EE = Ethiopia homozygous.