Integrating genomics and multivariate evolutionary quantitative genetics: a case study of constraints on sexual selection in Drosophila serrata

In evolutionary quantitative genetics, the genetic variance–covariance matrix, G , and the vector of directional selection gradients, β , are key parameters for predicting multivariate selection responses and genetic constraints. Historically, investigations of G and β have not overlapped with those dissecting the genetic basis of quantitative traits. Thus, it remains unknown whether these parameters reflect pleiotropic effects at individual loci. Here, we integrate multivariate genome-wide association study (GWAS) with G and β estimation in a well-studied system of multivariate constraint: sexual selection on male cuticular hydrocarbons (CHCs) in Drosophila serrata . In a panel of wild-derived re-sequenced lines, we augment genome-based restricted maximum likelihood to estimate G alongside multivariate single nucleotide polymorphism (SNP) effects, detecting 532 significant associations from 1 652 276 SNPs. Constraint was evident, with β lying in a direction of G with low evolvability. Interestingly, minor frequency alleles typically increased male CHC-attractiveness suggesting opposing natural selection on β . SNP effects were significantly misaligned with the major eigenvector of G , g max , but well aligned to the second and third eigenvectors g 2 and g 3 . We discuss potential factors leading to these varied results including multivariate stabilizing selection and mutational bias. Our framework may be useful as researchers increasingly access genomic methods to study multivariate selection responses in wild populations.

Specialization directs habitat selection responses to a top predator in semiaquatic but not aquatic taxa

Habitat selectivity has become an increasingly acknowledged mechanism shaping the structure of freshwater communities; however, most studies have focused on the effect of predators and competitors, neglecting habitat complexity and specialization. In this study, we examined the habitat selection of semiaquatic (amphibians: Bufonidae; odonates: Libellulidae) and aquatic organisms (true bugs: Notonectidae; diving beetles: Dytiscidae). From each family, we selected one habitat generalist species able to coexist with fish (Bufo bufo, Sympetrum sanguineum, Notonecta glauca, Dytiscus marginalis) and one species specialized in fishless habitats (Bufotes viridis, Sympetrum danae, Notonecta obliqua, Acilius sulcatus). In a mesocosm experiment, we quantified habitat selection decisions in response to the non-consumptive presence of fish (Carassius auratus) and vegetation structure mimicking different successional stages of aquatic habitats (no macrophytes; submerged and floating macrophytes; submerged, floating, and littoral-emergent macrophytes). No congruence between habitat specialists and generalists was observed, but a similar response to fish and vegetation structure defined both semiaquatic and aquatic organisms. While semiaquatic generalists did not distinguish between fish and fishless pools, specialists avoided fish-occupied pools and had a preferred vegetation structure. In aquatic taxa, predator presence affected habitat selection only in combination with vegetation structure, and all species preferred fishless pools with floating and submerged macrophytes. Fish presence triggered avoidance only in the generalist bug N. glauca. Our results highlight the significance of habitat selectivity for structuring freshwater ecosystems and illustrate how habitat selection responses to a top predator are dictated by specialization and life history.

Experimental Evolution of a Pheromone Signal

ABSTRACTSexual signals are important in speciation, but understanding their evolution is complex as these signals are often composed of multiple, genetically interdependent components. To understand how signals evolve, we thus need to consider selection responses in multiple components and account for the genetic correlations among components. One intriguing possibility is that selection changes genetic correlations between components under selection and those not under selection, reducing evolutionary constraints. However, this hypothesis remains largely untested empirically. In this study, we investigate the evolutionary response of the multi-component female sex pheromone blend of the moth Heliothis subflexa to 10 generations of artificial selection. We observed a selection response of about 3/4s of a phenotypic standard deviation in the components under selection. Interestingly, other pheromone components that are biochemically and genetically linked to the components under selection did not change. We found that this component-specific selection response was likely facilitated due to reduced genetic coupling between the component under selection and the components not under selection, resulting from changes in genetic covariances within the 10 generations of selection. Our findings provide rare empirical support for an intriguing mechanism by which a sexual signal can respond to selection without possible constraints from indirect selection responses.SUMMARY STATEMENTThis study reveals a mechanism that can facilitate evolution of sex pheromone components under selection independent of other components of the sex pheromone that are not under selection.

Predicted responses to selection across the climatic range of a rainforest Drosophila without local adaptation: environmental variation limits trait divergence along ecological gradients

Evolutionary responses to environmental change require heritable variation in traits under selection. Both heritability and selection vary with the environment, and may also covary, meaning that environmental variation can be an important source of evolutionary constraint. However, estimates of heritability and selection along environmental gradients in the field are rare. We estimated environmental variation in selection on three traits (cold tolerance, heat tolerance and wing size) of the rainforest fly Drosophila birchii by transplanting flies in cages along two elevational gradients in north-east Queensland, Australia, and calculating the genetic covariance of trait values with cage productivity at each elevation. We estimated heritability of each trait from laboratory crosses, and environmental variation in heritability of wing size from the correlation of mothers and daughters in cages at each elevation. We then used estimates of selection and heritability to predict selection responses along the elevation gradients. Laboratory assays revealed low-moderate genetic variation in all traits and low covariation among traits, suggesting the potential for a strong response to selection. Estimated selection responses predicted divergence of cold tolerance with elevation at one gradient. However, this was not observed at either gradient, with no difference between high and low elevation populations for this trait. Despite substantial variation in heritability (and predicted selection response) of wing size, this appeared random with respect to elevation, preventing overall divergence and suggesting that local environmental variation constrains evolutionary responses along natural ecological gradients. Such an effect, if widespread, may significantly slow evolutionary responses to environmental change.

Integrating genomics and multivariate evolutionary quantitative genetics: A case study of multivariate constraints on sexual selection in Drosophila serrata

In evolutionary quantitative genetics, the genetic variance-covariance matrix, G, and the vector of directional selection gradients, β , are key parameters for predicting multivariate selection responses and genetic constraints. Historically, investigations of G and β have not overlapped with those dissecting the genetic basis of quantitative traits. Thus, it remains unknown whether these parameters reflect pleiotropic effects at individual loci. Here, we integrate multivariate GWAS with G and β estimation in a well-studied system of multivariate constraint; sexual selection on male cuticular hydrocarbons (CHCs) in Drosophila serrata. In a panel of wild-derived resequenced lines, we augment genome-based REML, (GREML) to estimate G alongside multivariate SNP effects, detecting 532 significant associations from 1,652,276 SNPs. Constraint was evident, with β lying in a direction of G with low evolvability. Interestingly, minor frequency alleles typically increased male CHC-attractiveness suggesting opposing natural selection on β. SNP effects were significantly misaligned with the major eigenvector of G, gmax, but well aligned to the second and third eigenvectors g2 and g3. We discuss potential factors leading to these varied results including multivariate stabilising selection and mutational bias. Our framework may be useful as researchers increasingly access genomic methods to study multivariate selection responses in wild populations.

Interplay between extreme drift and selection intensities favors the fixation of beneficial mutations in selfing maize populations

Population and quantitative genetic models provide useful approximations to predict long-term selection responses sustaining phenotypic shifts, and underlying multilocus adaptive dynamics. Valid across a broad range of parameters, their use for understanding the adaptive dynamics of small selfing populations undergoing strong selection intensity (thereafter High Drift-High selection regime, HDHS) remains to be explored. Saclay Divergent Selection Experiments (DSEs) on maize flowering time provide an interesting example of populations evolving under HDHS, with significant selection responses over 20 generations in two directions. We combined experimental data from Saclay DSEs, forward individual-based simulations, and theoretical predictions to dissect the evolutionary mechanisms at play in the observed selection responses. We asked two main questions: How do mutations arise, spread, and reach fixation in populations evolving under HDHS? How does the interplay between drift and selection influence observed phenotypic shifts? We showed that the long-lasting response to selection in small populations is due to the rapid fixation of mutations occurring during the generations of selection. Among fixed mutations, we also found a clear signal of enrichment for beneficial mutations revealing a limited cost of selection. Both environmental stochasticity and variation in selection coefficients likely contributed to exacerbate mutational effects, thereby facilitating selection grasp and fixation of small-effect mutations. Together our results highlight that despite a small number of polymorphic loci expected under HDHS, adaptive variation is continuously fueled by a vast mutational target. We discuss our results in the context of breeding and long-term survival of small selfing populations.

Low-coverage sequencing in a deep intercross of the Virginia body weight lines provides insight to the polygenic genetic architecture of growth: novel loci revealed by increased power and improved genome-coverage

Genetic dissection of highly polygenic traits is a challenge, in part due to the power necessary to confidently identify loci with minor effects. Experimental crosses are valuable resources for mapping such traits. Traditionally, genome-wide analyses of experimental crosses have targeted major loci using data from a single generation, often the F2, with additional, later generation individuals being generated for replication and fine-mapping. Here, we aim to confidently identify minor-effect loci contributing to the highly polygenic basis of the long-term, divergent bi-directional selection responses for 56-day body weight in the Virginia chicken lines. To achieve this, a powerful strategy was developed to make use of data from all generations (F2-F18) of an advanced intercross line, developed by crossing the low and high selected lines after 40 generations of selection. A cost-efficient low-coverage sequencing based approach was used to obtain high-confidence genotypes in 1Mb bins across 99.3% of the chicken genome for >3,300 intercross individuals. In total, 12 genome-wide significant and 10 additional suggestive QTL for 56-day body weight were mapped, with only two of these QTL reaching genome-wide, and one suggestive, significance in analyses of the F2 generation. Five of the significant, and four of the suggestive, QTL were among the 20 loci reaching a 20% FDR-threshold in previous analyses of data from generation F15. The novel, minor-effect QTL mapped here were generally mapped due to an overall increase in power by integrating data across generations, with minor contributions from increased genome-coverage and improved marker information content. Significant and suggestive QTL now explain >60% of the difference between the parental lines, three times more than the previously reported significant QTL. Making integrated use of all available samples from multiple generations in experimental crosses is now economically feasible using the low-cost, sequencing-based genotyping strategies outlined here. Our empirical results illustrate the value of this strategy for mapping novel minor-effect loci contributing to complex traits to provide a more confident, comprehensive view of the individual loci that form the genetic basis of the highly polygenic, long-term selection responses for 56-day body weight in the Virginia chicken lines.