ANONYMISED Data and code for "Genetic variance and phenotypic selection on pathogen-linked oviposition choice in Drosophila"

This entry contains raw data collected during two experiments on pathogen avoidance and analysis code (R), as described below: Fly stocks We studied an outbred Drosophila melanogaster population (Ashworth Outcrossed, AOx) created by crossing 113 DGRP lines and then randomly outbred for at least 151 generations (at the time of the experiment) (Monteith et al. 2019, 2024; Savola et al. 2021). To generate mated focal females, 20 vials (sugar-cornmeal medium; Table S1) were set up with 10 females and 5 males. After 24 h adults were transferred to fresh vials for a further 24 h, then removed. Females emerging from eggs laid in these vials were used in choice assays. Thirty-six hours before trial, each focal female was placed with two males and yeast in a food vial to encourage mating. All flies were 2–5 days old, maintained at 25°C (±2°C) under 12:12 h light:dark and reared on sugar–cornmeal medium (Table S1). For heritability estimates, 24 randomly chosen DGRP lines (Mackay et al., 2012) were maintained at 18°C (±2°C), 12:12 h light:dark for two weeks. For each line, 10 females and 5 males were placed in a food vial and transferred to new vials every 3-4 days. Females aged 1-4 days were assayed. Thirty-six hours before each oviposition assay, each female was placed with two males from her line and yeast. Assay conditions were 25°C (±2°C), 12:12 h light:dark. Bacterial culturing All culturing was performed in a microbiological safety cabinet (MSC). Luria-Bertani (LB) broth (20 ml) was dispensed into two 50 ml tubes. One tube was inoculated with frozen Pseudomonas aeruginosa PA14 (−70°C stock) using a sterile loop; the other served as an uninoculated LB control. Tubes were loosely capped to allow aeration and taped. Cultures were incubated at 37°C with shaking at 120 rpm for 18 h to obtain an overnight culture. Choice chamber and oviposition assay Two plastic caps (Bijou vial caps) were filled with sugar-cornmeal medium (Table S1). One cap was overlaid with 100 µl of the PA14 overnight culture and the other with 100 µl of sterile LB (clean control). After drying in the MSC, caps were fixed 45 mm apart on opposite sides of a 90 mm lidded, Petri dish using BluTack to form a two-choice chamber. Control chambers contained two clean caps. A single mated female was very lightly anaesthetised with CO2 and placed centrally in the chamber. Dishes were sealed with tape and incubated for 24-28 h at 25°C (±2°C), 12:12 h light:dark. Start times were standardised (10:00-12:00), and the position of sites relative to the incubator walls was alternated across assays to distribute positional effects. Females were then removed from the assay chambers under CO2 anaesthesia and the number of eggs on each substrate were counted under a dissecting microscope. Oviposition index and fitness components The oviposition index (OI) was defined as OI = (C − B)/(C + B), where C and B are eggs laid on the Clean and Bacterial substrates, respectively. OI ranges from −1 (all eggs on bacteria) to 1 (all eggs on clean). For each female we recorded: total eggs laid; egg-to-adult viability (proportion of eggs eclosing as adults per site); total adult offspring. Given known sex differences in susceptibility to P. aeruginosa in adult D. melanogaster (Vincent and Sharp 2014; Gupta et al. 2017), we also tested whether the sex ratio of surviving offspring differed between substrates. To measure fly development, each cap was taped onto a standard vial containing fly medium and incubated at 25°C (±2°C), 12:12 h light:dark, until adult emergence (13–14 days). Adults from both sites were counted to calculate egg viability, total adult offspring and sex ratio. Sample sizes and exclusions Altogether, we assayed 130 single females: 100 in treatment chambers (clean versus bacterial substrate) and 30 in control chambers (two clean substrates). To avoid undue leverage of very small clutches, females laying fewer than 10 eggs were excluded from all analyses, yielding a final sample size of 87 treatment and 27 control females. For 15 treatment and 2 control females, ≤6 adults escaped; these were included in egg viability but excluded from sex ratio. Within-female contrasts between substrates (egg number, viability, offspring number, sex ratio) were thus based on paired data from the retained treatment females. We also analysed the subset of assays where females laid exclusively on one substrate (where OI = −1 or 1); sample sizes for these contrasts are given with the model outputs (Table 1, Models 5–8). For the heritability assay, 24 DGRP lines, with ten females per line, were tested using the same protocol described above (240 choice assays). Phenotypic selection and heritability We quantified phenotypic selection acting pathogen-related egg-laying choice by regressing the mean-standardised fitness components (egg number, viability, offspring number, sex ratio) on the oviposition index (OI) to estimate both linear (β) and quadratic (γ) selection gradients (Kingsolver and Pfennig 2007; Queller 2017; Walsh and Lynch 2018). Standardised fitness for each component was calculated by dividing the individual value by the corresponding population mean (W), such that values >1 indicate above-average performance. Broad-sense heritability (H2) of OI was estimated using ten females from each of 24 randomly selected DGRP lines. We calculated H2 = VG/(VG + VE), where VG is the among-line variance in OI. Because OI was modelled as a binomial trait, its environmental variance VE is the sum of the residual variance VR and the variance of the logistic distribution (Nakagawa and Schielzeth 2010; Mackay and Huang 2018). The within-line (environmental) variance VE is therefore calculated as VR+ Statistical analysis Analyses were conducted in R 4.3.3 (R Core Team, 2023) using ggplot2 (Wickham 2009), lme4 (Bates et al. 2015) and lmerTest (Kuznetsova et al. 2017). Linear mixed models (LMMs) were used for egg number, adult offspring number, OI and relative fitness components when residuals were approximately normal; generalised linear mixed models (GLMMs) were used otherwise. Experimental block was fitted as a random effect in all mixed models. For within-individual comparisons across substrates (egg number, viability, offspring number, sex ratio), fly identity was also fitted as a random effect. Offspring counts per site were analysed with Poisson GLMMs; egg-to-adult viability and sex ratio with binomial GLMMs using cbind(successes, failures) in order to analyse the number of eggs, rather than the ratio. We compared bacterial versus clean substrates, treatment versus control conditions, within-control differences, and the two extreme OI classes (−1 versus 1). For heritability, a binomial GLMM with cbind(eggs on clean, eggs on bacteria) as the response and DGRP line as a random effect was fitted; among-line and residual variances were extracted to compute H2 as described above.