Skip to main content
Dryad logo

Hormonal pleiotropy structures genetic covariance

Citation

Wittman, Tyler; Robinson, Christopher; McGlothlin, Joel; Cox, Robert (2021), Hormonal pleiotropy structures genetic covariance, Dryad, Dataset, https://doi.org/10.5061/dryad.1rn8pk0tf

Abstract

Quantitative genetic theory proposes that phenotypic evolution is shaped by G, the matrix of genetic variances and covariances among traits. In species with separate sexes, the evolution of sexual dimorphism is also shaped by B, the matrix of between-sex genetic variances and covariances. Despite considerable focus on estimating these matrices, their underlying biological mechanisms are largely speculative. We experimentally tested the hypothesis that G and B are structured by hormonal pleiotropy, which occurs when one hormone influences multiple phenotypes. Using juvenile brown anole lizards (Anolis sagrei) bred in a paternal half-sibling design, we elevated the steroid hormone testosterone with slow-release implants while administering empty implants to siblings as a control. We quantified effects of this manipulation on the genetic architecture of a suite of sexually dimorphic traits, including body size (males are larger than females) and the area, hue, saturation, and brightness of the dewlap (a colorful ornament that is larger in males than in females). Testosterone masculinized females by increasing body size and dewlap area, hue, and saturation, while reducing dewlap brightness. Control females and males differed significantly in G, but treatment of females with testosterone rendered G statistically indistinguishable from males. Whereas B was characterized by low between-sex genetic correlations when estimated between control females and control males, these same correlations increased significantly when estimated between testosterone females and either control or testosterone males. The full G matrix (including B) for testosterone females and either control or testosterone males was significantly less permissive of sexually dimorphic evolution than was G estimated between control females and control males, suggesting that natural sex differences in testosterone help decouple genetic variance between the sexes. Our results confirm that hormonal pleiotropy structures genetic covariance, implying that hormones play an important yet overlooked role in mediating evolutionary responses to selection.

Methods

We experimentally tested the hypothesis that G and B are structured by hormonal pleiotropy, which occurs when one hormone influences multiple phenotypes. We bred anoles in a paternal half-sibling design with two dams per sire (n = 120 dams, 60 sires). We raised F3 progeny to 3 months of age and then administered one of two treatments: (1) a slow-release implant containing 100 μg testosterone, or (2) an empty implant as a control. To balance treatments within maternal families, we haphazardly determined whether the first offspring of each sex would receive a testosterone or a control implant for a given family, then alternated between treatment groups for all subsequent progeny of each sex. At 8 months of age, we measured each individual for snout-vent length (length between the tip of the snout and the cloaca, measured to the nearest mm) and photographed its dewlap to measure area, hue, saturation, and brightness. We uploaded images into ImageJ and set the scale of measurement using the 5-mm grids of the graph paper. We measured area (mm2) by outlining the dewlap from its anterior projection from chin to its posterior attachment to the venter using the “polygon” tool. To quantify the color of the dewlap, we used the “oval” tool to define a circle in the center of the dewlap, with diameter of the circle equal to 1/3 the width of the dewlap, providing a consistent measure of the “center” of each dewlap despite variation in its absolute size. We used the “color histogram” function to extract the mean red, green, and blue values for the selected area, then transformed these values into hue (primary color reflected, measured on a 360° color wheel), saturation (purity of the hue, 0% = achromatic, 100% = pure color), and brightness (relative to maximum possible for color of the same hue and saturation, 0% = black, 100% = white-tint-pure color) using the rgb2hsv function of the package gDevices within R3.6.2 (R Core Team 2019). To quantify sexual dimorphism before treatment we measured these same traits following the same procedure on a subset of individuals at three months of age. We estimated genetic covariance matricies (G) for these five traits using the program WOMBAT, and a pedigree describing the relationships among F1 grandparents, F2 parents, and F3 experimental progeny. We estimated separate within-sex G matrices for each of the four experimental groups (control females, control males, testosterone females, testosterone males). For pairs of male and female treatments, we estimated full G matrices including both within-sex matrices (GF, GM) and the between-sex matrix (B). We standardized the genetic covariance matrices to the phenotypic variance. We performed analysis on both standardized and unstandardized matricies. 

Usage Notes

We provide all the files necessary to preform the analysis presented in the manuscript, as well as the data files and pedigree used to estimate the genetic covariance matrices. Readme files detailing the types of data given in each file and how to use them are provided. Further, code used for the analysis presented in the manuscript is provided at https://github.com/ty-wittman/evo_qg_analysis_r_code.