Data and code from: Longevity hinders evolutionary rescue through slower growth but not necessarily slower adaptation
Data files
Oct 16, 2025 version files 324.45 MB
-
README.md
8.59 KB
-
sim_results_m1_ages_equal-lambda.csv
143.10 MB
-
sim_results_m1_ages.csv
135.90 MB
-
sim_results_m1_allsizes_n_equal-lambda.csv
108.41 KB
-
sim_results_m1_allsizes_n_equalg20.csv
108.14 KB
-
sim_results_m1_allsizes_n.csv
108.93 KB
-
sim_results_m1_disaggregated_n_equal-lambda.csv
867.18 KB
-
sim_results_m1_disaggregated_n_equalg20.csv
1.08 MB
-
sim_results_m1_disaggregated_n.csv
1.15 MB
-
sim_results_m1_phenotypic_components_equal-lambda.csv
11.42 MB
-
sim_results_m1_phtype_equalg20.csv
15.14 MB
-
sim_results_m1_phtype.csv
15.46 MB
Abstract
''Evolutionary rescue'' is the process by which populations experiencing severe environmental change avoid extinction through adaptation. Applying theory to natural populations and conservation targets requires investigating the effects of several life-history traits, including longevity. Theory demonstrates that longevity can inhibit rescue through slower phenotypic adaptation when selection acts once per lifetime, leaving open questions about longevity's effects when individuals face multiple rounds of selection. We developed a model integrating evolutionary rescue with concepts from life-history theory, particularly the trade-off where increasing longevity produces slower population growth rates. Our model varies longevity by modifying the balance of survival and reproduction, with selection acting on survival allowing for adaptation within cohorts. We used this model to study life-history strategies with different longevities responding to sudden environmental change. Simulations demonstrated that higher longevity resulted in more time at low density and increased extinction. With perfect trait heritability, rates of adaptation were nearly identical across longevities. But at lower heritabilities, repeated selection under longevity decoupled mean population phenotypes and genotypes, producing a transient phase of rapid phenotypic change. Our results demonstrate that longevity impedes rescue by slowing population growth but does not always slow rates of adaptation.
All simulation code (in language R, using packages dplyr, tidyr, mc2, and parallel for code and ggplot2 and cowplot for figures) and output (in .csv form) for paper "Longevity hinders evolutionary rescue through slower growth but not necessarily slower adaptation".
Contents include:
- Functions for running simulations
- Scripts to run simulated experiments (including outputs)
- Scripts for validating simulation outputs (including figures)
- Scripts for generating figures in manuscripts
Main simulation analysis output files
File sim_results_m1_allsizes_n.csv:
Mean and variance of population size for each of 27 parameter combinations over 101 time steps for main batch of simulations. One row corresponds to a summary of one time step within a parameter combination. This file was generated by script run_sims/model1_popsize_extinctions.R and used to generate Figures 2 and 3 in main text.
Fields:
s.max- maximum survival value ($\widehat{s}$ in paper)var.z- initial population-level phenotypic variance ($\gamma^2$ in paper)h2- heritability ($h^2$ in paper)t- time stepnbar- mean of population size across simulation trials with given parameter combination in time stepnvar- variance of population size across simulation trials with given parameter combination in time steppsrv- proportion of surviving populations (i.e., with size > 0) within parameter combination at the given time stepnn- number of trials used to estimate mean and variance
File sim_results_m1_disaggregated_n.csv:
Simulation output (recording size only) for 20 trials individual trials per parameter combination over 101 time steps (or time of extinction, whichever occurs first) for each of 27 parameter combinations in "main" batch of simulations. One row corresponds to one time step within one simulation trial. This file was generated by script run_sims/model1_popsize_extinctions.R and used to generate Figure 2 in main text.
Fields:
t- time stepn- population size in time steptrial- simulation trial numbers.max- maximum survival value ($\widehat{s}$ in paper)var.z- phenotypic variance ($\gamma^2$ in paper)h2- heritability ($h^2$ in paper)
File sim_results_m1_phtype.csv:
Mean and variance of each of three phenotypic components (breeding value, environmental component of phenotype, and phenotype) for 200 simulated population over time for nine parameter combinations over 51 time steps. One row corresponds to one time step within one simulation trial. This file was generated by script run_sims/model1_phenotypic_components.R and used to generate Figures 4 and 6 in main text and Fig. S28 in Supporting Info.
Fields:
t- time stepn- population size in time step (note: not used in analysis)bbar- mean breeding value in population within time step ($\bar{b}$ in paper)bvar- variance of breeding values in population within time step ($\gamma^2_a$ in paper)bvarw- not used in analysisebar- mean environmental component of phenotype in population within time step ($\bar{e}$ in paper)evar- variance of environmental component of phenotype in population within time step ($\gamma^2_e$ in paper)evarw- not used in analysiszbar- mean phenotype in population within time step ($\bar{z}$ in paper)zvar- variance of phenotypes in population within time step ($\gamma^2$ in paper)zvarw- not used in analysistrial- simulation trial numbers.max- maximum survival value ($\widehat{s}$ in paper)h2- heritability ($h^2$ in paper)
File sim_results_m1_ages.csv:
Mean and variance of each of three phenotypic components (breeding value, environmental component of phenotype, and phenotype) and size of each cohort (age class) for 200 simulated population over time for nine parameter combinations over 51 time steps. One row corresponds to one cohort within time step within one simulation trial. This file was generated by script run_sims/model1_age_structure.R and used to generate Figure 5 in main text and Fig. S27 in Supporting Info.
Fields:
t- time stepage- age of cohort (denoted $k$ in paper)n- cohort size in time stepr- number of offspring sired by cohort (note: not used in analysis)bbar- mean breeding value in cohort within time step ($\bar{b}$ in paper) (note: not used in analysis)bvar- variance of breeding values in cohort within time step ($\gamma^2_{a,k}$ in paper) (note: not used in analysis)ebar- mean environmental component of phenotype in cohort within time step ($\bar{e}_k$ in paper) (note: not used in analysis)evar- variance of environmental component of phenotype in cohort within time step ($\gamma^2_{e,k}$ in paper) (note: not used in analysis)zbar- mean phenotype in cohort within timestep ($\bar{z}_k$ in paper) (note: not used in analysis)zvar- variance of phenotypes in cohort within time step ($\gamma^2_k$ in paper) (note: not used in analysis)trial- simulation trial numbers.max- maximum survival value ($\widehat{s}$ in paper)h2- heritability ($h^2$ in paper)
Additional simulations
Equal-$lambda$ simulations
These additional simulations were run with all life-history strategies having equal maximum per-time step growth rates rather than maximum intrinsic fitness equivalent (as in main simulations). Files are named and largely structured similarly to the main output files described above but with the following exceptions:
Files sim_results_m1_allsizes_n_equal-lambda.csv and sim_results_m1_disaggregated_n_equal-lambda.csv:
This file includes summaries of only 500 trials per combination. These were created by script run_sims/equal_lambda/model1_popsize_extinctions_lambda.R and were used to make Figs. S17 and S18 in Supporting Information.
Files sim_results_m1_phenotypic_components_equal-lambda.csv and sim_results_m1_ages_equal-lambda.csv:
These files recorded the equilibrium population size (lstar, equivalent to $\lambda^*$ in main text) of each life-history strategy. The phenotypic component file includes the s.max used in the treatment, but by mistake the age structure file does not. The longevity treatments are assigned using the following R code:
n.all = mutate(
n.all,
long = cut(lstar, breaks = c(0, 1.185, 1.19, Inf), labels = c('low', 'medium', 'high'))
)
These files were generated, respectively, by scripts run_sims/equal_lambda/model1_phenotypic_components_lambda.R and run_sims/equal_lambda/model1_age_structure_lambda.R and were used to make Figures S20-22.
Equal-$\gamma^2_0$ simulations
These additional simulations were run with all life-history strategies having equal phenotypic variance in newborn cohorts rather than population-level phenotypic variance equivalent (as in main simulations). Files are named and largely structured similarly to the main output files described above but with the following exceptions:
Files sim_results_m1_allsizes_n_equalg20.csv and sim_results_m1_disaggregated_n_equalg20.csv:
Note that here the variable var.z corresponds to the phenotypic variance in a newborn cohort (rather than at the population level). These files were created by the script run_sims/model1_popsize_extinctions_equal_g20.R and were used to make Figs. S23 and S24 in Supporting Information.
File sim_results_m1_phtype_equalg20.csv:
This file was created by the script run_sims/model1_phenotpyic_components_equal_g20.R and was used to make Figures S25 and S26 in Supporting Information.
Note on variances "missing" data (NA in fields)
All variances reported in these files are sample variances rather than population variances. NA appears in some files in columns reporting variances (bvar, evar, and zvar) where the sample size is one (i.e., n is equal to 1 for the corresponding row in the file). This is because variance reported is sample variance, under which conditions variance is undefined and the R function var() function returns NA.
Scripts
All code to re-create analysis, including but not limited to data files presented here, are publicly available on GitHub at the address https://github.com/melbourne-lab/evo_rescue_longevity. Please contact scottwatsonnordstrom@gmail.com with questions.