Sexual selection matters in genetic rescue, but productivity benefits fade over time: A multi-generation experiment to inform conservation
Data files
Dec 17, 2024 version files 26.46 KB
-
README.md
4.40 KB
-
SEL-GR.csv
10.16 KB
-
SS-GR.csv
5.83 KB
-
Stress-SEL-GR.csv
2.80 KB
-
Stress-SS-GR.csv
3.28 KB
Abstract
Globally, many species are threatened by population decline because of anthropogenic changes leading to population fragmentation, genetic isolation, and inbreeding depression. Genetic rescue, the controlled introduction of genetic variation, is a method used to relieve such effects in small populations. However, without understanding how the characteristics of rescuers impact rescue attempts interventions run the risk of being sub-optimal, or even counterproductive. We use the Red Flour Beetle (Tribolium castaneum) to test the impact of rescuer sex, and sexual selection background, on population productivity. We record the impact of genetic rescue on population productivity in 24 and 36 replicated populations for ten generations following intervention. We find little or no impact of rescuer sex on the efficacy of rescue but show that a background of elevated sexual selection makes individuals more effective rescuers. In both experiments, rescue effects diminish 6-10 generations after the rescue. Our results confirm that the efficacy of genetic rescue can be influenced by characteristics of the rescuers and that the level of sexual selection in the rescuing population is an important factor. We show that any increase in fitness associated with rescue may last for a limited number of generations, suggesting implications for conservation policy and practice.
README: Sexual selection matters in genetic rescue, but productivity benefits fade over time: A multi-generation experiment to inform conservation
https://doi.org/10.5061/dryad.g1jwstr1f
Description of the data and file structure
Test of sex and, sexual selection background, of genetic rescuers on the fitness of an inbred population.
Null entries in the data sets were not included in that or any future generations.
Abbreviations
KSS - Krakow Super Strain, Outbred ancestral population.
C - Control treatment, Inbred populations that received no rescue.
F - Female rescue treatment, Inbred populations that received rescue by a female KSS.
M - Male rescue treatment, Inbred populations that received rescue by a male KSS.
Control - Control treatment, Inbred populations that received no rescue.
Mo - Monogamy rescue treatment, Inbred populations that received rescue by a monogamous beetle.
Po - Polyandry rescue treatment, Inbred populations that received rescue by a Polyandrous beetle.
Mo control - Control population of monogamous beetles in experimental setup conditions (Not included in the analysis).
Po control - Control population of polygamous beetles in experimental setup conditions (Not included in the analysis).
Files and variables
File: SEL-GR.csv
Description: Productivity data for experiment with genetic rescue by different sexual selection regimes
Variables
- ID: Identifying number
- Generation: Generation after rescue
- Productivity: Number of adult offspring produced
- Isoline: Inbred line Identification
- Treatment: Treatment applied
File: SS-GR.csv
Description: Productivity data for experiment with genetic rescue by different sexes
Variables
- ID: Identifying number
- Generation: Generation after rescue
- Productivity: Number of adult offspring produced
- Isoline: Inbred line Identification
- Treatment: Treatment applied
File: Stress-SEL-GR.csv
Description: Productivity data for experiment with genetic rescue by different sexual selection regimes in stressful conditions
Variables
- ID: Identifying number
- Generation: Generation after rescue
- Productivity: Number of adult offspring produced
- Isoline: Inbred line Identification
- Treatment: Treatment applied
File: Stress-SS-GR.csv
Description: Productivity data for experiment with genetic rescue by different sexes in stressful conditions
Variables
- ID: Identifying number
- Generation: Generation after rescue
- Productivity: Number of adult offspring produced
- Isoline: Inbred line Identification
- Treatment: Treatment applied
Code/software
Statistical analyses were carried out in R V4.4.1 (R Core Team, 2024) utilising R studio version 2024.04.2+764 (Posit team, 2024). Tidyverse (Wickham et al., 2019), stats (R Core Team, 2024), Rmisc (Hope, 2022)and googlesheets4 (Bryan, 2023) were used for data management and exploration. Plots were created using ggplot2 (Wickham, 2016). The distribution of data was checked using the shapiro.test function (R Core Team, 2024). Generalised Linear Mixed Models (GLMMs) were fitted to test for differences in productivity between the experimental treatments using glmmTMB (Brooks et al., 2017). Model fit was checked using DHARMa (Hartig, 2022). Model parameters were checked for collinearity using variance inflation factor (Vif) scores with the check_collinearity function from performance (Lüdecke et al., 2021). There were no issues with overdispersion or collinearity (VIF: <3 for all variables) in any models. R2 was determined using the r.squaredGLMM function in MuMIn (Bartoń, 2024). Post-hoc pairwise Tukey tests were carried out using multcomp (Hothorn et al., 2008).
full_script - R script as one whole file.
SEL-GR LOAD - Load packages and data for sexual selection experiments
SEL-GR FIGURE - Create figure for sexual selection experiments
SEL-GR MODELS - Run models for sexual selection experiments
SS-GR LOAD - Load packages and data for sex experiments
SS-GR FIGURE - Create figure for sex experiments
SS-GR MODELS - Run models for sex experiments
Supplementary files
README.pdf - PDF version of the README
Figure 1 - Figure 1 from the manuscript
Figure 2 - Figure 2 from the manuscript
Figure 3 - Figure 3 from the manuscript
Figure 4 - Figure 4 from the manuscript
Methods
Husbandry
T. castaneum were kept in a controlled environment at 30°C and 60% humidity with a 12:12 light-dark cycle. Populations were kept on standard fodder consisting of 90% organic white flour, 10% brewer’s yeast and a layer of oats for traction unless otherwise stated. During the husbandry cycle, 2mm and 850µm sieves were used to remove pupae and adults from fodder. The following cycle was started by a set number of adults (line dependent, see below) being placed into containers with fresh standard fodder. The oviposition phase: populations were given seven days to mate and lay eggs before adults were removed by sieving to prevent overlapping generations. The fodder containing eggs was returned to the container. The development phase: eggs were kept in the containers for 35 days to allow the eggs to develop into mature adults. Around day 21 of the development phase, pupae were collected to obtain known-sex virgin individuals which were then used to start the next generation. The pupae were kept as virgins in single-sex groups of 20 for 10 days to allow them to complete development. Once mature, the cycle began again with those beetles going into fresh fodder to form a population of males and females.
Tribolium castaneum lines
Krakow Super Strain (KSS): was created by mixing fourteen laboratory strains to maximise genetic diversity in a single strain (Laskowski et al., 2015). This was used as the outbred treatment in the genetic rescue experiments.
Inbred Lines: Founded from KSS and inbred through three single-pair bottlenecks in the first, fifth and seventh generations. Between bottlenecks, the lines were maintained at a maximum population size of 100 randomly selected adults. Of the initial 30 lines, 24 survived the inbreeding treatment and 12 lines were maintained and used for experiments.
Sexual Selection Lines: polyandrous and monogamous lines were created from the Georgia 1 stock (Haliscak and Beeman, 1983; Lumley et al., 2015). Each polyandrous line (n=3) was maintained each generation in twelve groups each consisting of five males and one female. Following oviposition, the eggs from all groups in a line are mixed to form one population from which the next generation’s groups will be sourced. For each monogamous line (n=3) twenty separate mating pairs are bred. Following oviposition, the eggs from all pairs are mixed and the next pairs are sourced from this population to maintain that line. The number of groups and pairs in each regime results in a theoretical Ne = 40 in each treatment (Godwin et al., 2020). These regimes had been maintained for 150 generations when rescuers were taken. The polyandrous lines are hereafter referred to as sexual selection lines, and monogamous as no sexual selection.
Genetic rescue protocol
Replicate experimental inbred populations were created from the inbred lines to serve as populations to be rescued. Pupae were sexed and placed into plastic dishes with lids, containing 10ml standard fodder in single-sex groups. 10±2 days after eclosion, ten males and ten females from a given line were placed in a 125ml tub with 70ml of standard fodder creating populations each containing twenty adult beetles at a 1:1 sex ratio for the oviposition phase. On day 20±1 of the development phase, pupae were again taken from the populations using the method outlined above to create the next non-overlapping generation.
Populations were maintained using twenty reproducing adults per generation, not allowing population growth. This allowed us to maintain a roughly constant population density during offspring development across generations, avoiding the confounding influence of negative density-dependence on offspring production (Duval et al., 1939; King and Dawson, 1972; Janus, 1989).
Each experimental population was randomly assigned an ID number, to avoid bias when handling. After being established at the experimental size, the populations were maintained in experimental conditions for one generation to avoid transgenerational density effects affecting the genetic rescue results (Đukić et al., 2021). The rescue treatments were applied in the second generation under experimental conditions. In each population, a single beetle was replaced with a rescuer thus maintaining the 1:1 sex ratio and population size, avoiding any increase in productivity due to a demographic rescue. Rescuers taken from their source populations as pupae were age-matched as closely as possible to individuals in experimental populations. On day 37 of the development phase experimental populations were frozen at -6°C and mature offspring were counted as a measure of productivity (our metric for population fitness). If a population was removed from the experiment because of slow development (pupae were not available to establish the next generation), that population was analysed as part of all generations prior but excluded henceforth.
The sex of the rescuer in genetic rescue
Due to logistic issues with ventilation, four out of the 12 experimental inbred populations failed to produce offspring in generation 0. From each of the remaining eight inbred lines, three replicate populations were created and assigned to one of three treatments; No Rescue control (ten inbred line males, ten inbred line females); Male Rescue (nine inbred line males, one KSS male, ten inbred line females); and Female Rescue (ten inbred line males, nine inbred line females, one KSS female; Figure 1). Populations were maintained for ten, non-overlapping generations.
Sexual selection and genetic rescue
We investigated the impact of a rescuer’s sexual selection history on the effectiveness of genetic rescue. From 12 inbred lines, three replicate populations were created and assigned to one of three treatments; No Rescue Control (ten inbred line males, ten inbred line females); Sexual Selection Rescue (nine inbred line males, one polyandrous male and, ten inbred line females); No Sexual Selection Rescue (nine inbred line males, one monogamous male, ten inbred line females; Figure 2). A single polyandrous and single monandrous line were used as the source for rescuers. Populations were maintained for nine generations.
Stressful conditions
To test rescue under stress conditions, duplicate rescue populations were established from each rescued line at generation five in the ‘sex’ experiment, and generation six in the ‘sexual selection’ experiment. These were maintained as in the main experiments (until generation ten and nine respectively), but with a reduction in the yeast content of the fodder, which is the main source of protein for the experimental populations. This reduction generates nutrient stress in T. castaneum (Godwin et al., 2020). In the ‘sex’ experiment fodder contained 0% yeast and 1% yeast in the ‘sexual selection’ experiment (because of low survival with zero yeast).
Statistical analyses
Statistical analyses were carried out in R V4.4.1 (R Core Team, 2024) utilising R studio version 2024.04.2+764 (Posit team, 2024). Tidyverse (Wickham et al., 2019), stats (R Core Team, 2024), Rmisc (Hope, 2022)and googlesheets4 (Bryan, 2023) were used for data management and exploration. Plots were created using ggplot2 (Wickham, 2016). The distribution of data was checked using the shapiro.test function (R Core Team, 2024). Generalised Linear Mixed Models (GLMMs) were fitted to test for differences in productivity between the experimental treatments using glmmTMB (Brooks et al., 2017). Model fit was checked using DHARMa (Hartig, 2022). Model parameters were checked for collinearity using variance inflation factor (Vif) scores with the check_collinearity function from performance (Lüdecke et al., 2021). There were no issues with overdispersion or collinearity (VIF: <3 for all variables) in any models. R2 was determined using the r.squaredGLMM function in MuMIn (Bartoń, 2024). Post-hoc pairwise Tukey tests were carried out using multcomp (Hothorn et al., 2008).
Within each experiment, we fitted GLMMs with the same model structure, using a negative binomial distribution to model productivity counts, which provided better model fit than a Poisson distribution. Productivity was the response variable, with treatment, generation and generation2 as fixed effects. Inbred line of origin and experimental population ID were included as random effects, with ID nested within inbred line. Interaction terms (treatment x generation, treatment x generation2) were initially included but removed from the model if not significant. The generation2 factor was not significant in the models for populations under stressful conditions and was therefore removed. When a quadratic effect of generation was found in a model, we then tested to see if this was due to an increase and a decrease in productivity, rather than just one of these. We did this using two separate GLMMs (with the same factors as previously) run on the data split into generations 1-5 and 5-10. These GLMMs were fitted with treatment and generation as fixed effects, ID nested within inbred line as a random effect. GLMMs were also fitted on generations 2 and 3 individually (Table S3 + S4) in the ‘sex’ experiment, these single-generation models used a Poisson distribution, productivity as a response variable and treatment as a fixed effect. Random effects were the same as above. This was to test at which point the rescue treatments resulted in a significant difference from the control, to see if there were differences in the speed of male or female rescue.