Diet can alter the cost of resistance to a natural parasite in Caenorhabditis elegans
Cite this dataset
Jiranek, Juliana; Gibson, Amanda (2023). Diet can alter the cost of resistance to a natural parasite in Caenorhabditis elegans [Dataset]. Dryad. https://doi.org/10.5061/dryad.2v6wwpzsv
Resistance to parasites confers a fitness advantage, yet hosts show substantial variation in resistance in natural populations. Evolutionary theory indicates that resistant and susceptible genotypes can coexist if resistance is costly, but there is mixed evidence that resistant individuals have lower fitness in the absence of parasites. One explanation for this discrepancy is that the cost of resistance varies with environmental context. We tested this hypothesis using Caenorhabditis elegans and its natural microsporidian parasite, Nematocida ironsii. We used multiple metrics to compare the fitness of two near-isogenic host genotypes differing at regions associated with resistance to N. ironsii. To quantify the effect of the environment on the cost associated with these known resistance regions, we measured fitness on three microbial diets. We found that the cost of resistance varied with both diet and the measure of fitness. We detected no cost to resistance, irrespective of diet, when fitness was measured as fecundity. However, we detected a cost when fitness was measured in terms of population growth, and the magnitude of this cost varied with diet. These results provide a proof-of-concept that, by mediating the cost of resistance, environmental context may govern the rate and nature of resistance evolution in heterogeneous environments.
We conducted two assays to assess different components of fitness across diet conditions. First, to determine if the resistant genotype has lower fecundity than the susceptible genotype, we counted the number of offspring produced by ERT250 and N2 hosts raised on the three bacterial diets. For each genotype, we isolated eggs using a standard bleach wash (Porta-de-la-Riva et al., 2012), then added approximately 100 eggs per plate to DA837-, OP50-, or HB101-seeded 100 mm plates. We allowed hosts to reach the fourth larval (L4) stage at 20°C then moved hosts individually to 35 mm plates seeded with the same bacteria as their original plate (n = 60 hosts, with one host per 35 mm plate; 10-14 hosts per genotype*food combination). We moved each host to a new plate every day for six days, at which point reproduction had finished. Plates with eggs were incubated at 20°C for 24 hours to hatch, then we counted viable offspring. We censored data from hosts that were lost or suffered damage unrelated to the test conditions (n = 4). Data from an additional three hosts were excluded from analysis because of reporting errors.
Population Expansion Assay
To test whether resistant populations grow at a slower rate than susceptible populations, we measured the size of resistant and susceptible populations after a standardized period of expansion. To ensure sufficient replication for each treatment, we limited this assay to DA837 and HB101. For both N2 and ERT250, we picked L4 hosts individually onto 100 mm plates seeded with either DA837 or HB101 (n = 100 hosts, with one host per 100 mm plate; 25 hosts per genotype*food combination). Plates were monitored daily. The populations expanded for six days (~2-3 generations), at which point we collected entire populations in M9 buffer, diluted the populations to 14.5 mL, and counted six 20 uL aliquots for each population. The total population size could be estimated by multiplying these aliquot counts by 725. Replicate populations were removed from the analysis if the original host died before producing offspring (n = 2) or if there was an error during the washing and resuspension steps (n = 4).
We performed all analyses in R Version 1.4.1106 (R Core Team, 2021). To test for variation in fitness, we used the lme4 package (Bates et al., 2015) to fit generalized linear models or mixed models (GLMM). For the fecundity data, we analyzed both variation in lifetime fecundity (the total number of offspring summed over all days) and variation in the number of offspring per day, which can reveal differences in reproductive timing. For the fecundity models, we found evidence of overdispersion under the Poisson distribution, so we assumed a negative binomial distribution. We compared multiple candidate models including host genotype, food type, and their interaction as predictors. For the daily number of offspring, we also included the day of observation and its interaction with each of the aforementioned predictors, as well as a random effect to account for repeated measures of the same host. For the population expansion data, we fit Poisson GLMMs, including host genotype, food type, and their interaction as predictors of hosts per aliquot, plus a random effect for the replicate population to account for repeated measures. For all analyses, we used AIC scores to compare candidate models.
National Institute of General Medical Sciences, Award: R35 GM137975-01
Bank of America