Last updated: 2019-06-18

Checks: 6 0

Knit directory: fitnessCostSD/

This reproducible R Markdown analysis was created with workflowr (version 1.3.0). The Checks tab describes the reproducibility checks that were applied when the results were created. The Past versions tab lists the development history.


Great! Since the R Markdown file has been committed to the Git repository, you know the exact version of the code that produced these results.

Great job! The global environment was empty. Objects defined in the global environment can affect the analysis in your R Markdown file in unknown ways. For reproduciblity it’s best to always run the code in an empty environment.

The command set.seed(20190312) was run prior to running the code in the R Markdown file. Setting a seed ensures that any results that rely on randomness, e.g. subsampling or permutations, are reproducible.

Great job! Recording the operating system, R version, and package versions is critical for reproducibility.

Nice! There were no cached chunks for this analysis, so you can be confident that you successfully produced the results during this run.

Great! You are using Git for version control. Tracking code development and connecting the code version to the results is critical for reproducibility. The version displayed above was the version of the Git repository at the time these results were generated.

Note that you need to be careful to ensure that all relevant files for the analysis have been committed to Git prior to generating the results (you can use wflow_publish or wflow_git_commit). workflowr only checks the R Markdown file, but you know if there are other scripts or data files that it depends on. Below is the status of the Git repository when the results were generated:


Ignored files:
    Ignored:    .DS_Store
    Ignored:    .Rhistory
    Ignored:    .Rproj.user/
    Ignored:    data/.DS_Store
    Ignored:    data/SD_k tests_2018_04_05.xlsx
    Ignored:    data/clean_data/.DS_Store
    Ignored:    data/data collection sheet - follow up looking at sex ratio.xlsx
    Ignored:    data/data collection sheet from Heidi.xlsx
    Ignored:    data/model_output/
    Ignored:    docs/.DS_Store
    Ignored:    docs/figure/SD_costs_analysis.Rmd/
    Ignored:    docs/figure/evolutionary_simulation_SD.Rmd/
    Ignored:    manuscript/.DS_Store
    Ignored:    output/

Unstaged changes:
    Modified:   figures/fig3.pdf

Note that any generated files, e.g. HTML, png, CSS, etc., are not included in this status report because it is ok for generated content to have uncommitted changes.


These are the previous versions of the R Markdown and HTML files. If you’ve configured a remote Git repository (see ?wflow_git_remote), click on the hyperlinks in the table below to view them.

File Version Author Date Message
Rmd 236bdf8 lukeholman 2019-06-18 Simulation redone
html 3244bce lukeholman 2019-06-11 Build site.
Rmd 7950fbb lukeholman 2019-06-11 Changes after Tom review
html 7950fbb lukeholman 2019-06-11 Changes after Tom review
html 0648e58 lukeholman 2019-06-11 Build site.
Rmd a85b1fc lukeholman 2019-06-11 Changes after Tom review
html 437b6d7 lukeholman 2019-06-11 Build site.
Rmd 472fa8d lukeholman 2019-06-11 Changes after Tom review
html e3cfdc4 lukeholman 2019-03-12 Build site.
Rmd e8a0b18 lukeholman 2019-03-12 fix
html 861ee2d lukeholman 2019-03-12 Build site.
html 6a50b04 lukeholman 2019-03-12 Build site.
Rmd b92f254 lukeholman 2019-03-12 added raw data
html b92f254 lukeholman 2019-03-12 added raw data
html 9f20dc5 lukeholman 2019-03-12 Build site.
Rmd 555801a lukeholman 2019-03-12 Fixed up supplement
html 555801a lukeholman 2019-03-12 Fixed up supplement
html 36e95d3 lukeholman 2019-03-12 Build site.
Rmd 8d92589 lukeholman 2019-03-12 added supplment
html 8d92589 lukeholman 2019-03-12 added supplment
html e67c963 lukeholman 2019-03-12 Build site.
Rmd cceac7a lukeholman 2019-03-12 Tweak
html ba43211 lukeholman 2019-03-12 Build site.
html 6b03f9c lukeholman 2019-03-12 Build site.
Rmd 2bc25a0 lukeholman 2019-03-12 Added main pages

Set up for the analysis

Load R packages

packages <- c("dplyr", "brms", "ggplot2", "reshape2", "Cairo", "knitr", "pander", "lazerhawk",
              "grid", "gridExtra", "ggthemes", "readr", "tibble", "stringr", "gameofthrones")
shh <- suppressMessages(lapply(packages, library, character.only = TRUE, quietly = TRUE))
nCores <- 1

summarise_brms <- function(brmsfit){
  lazerhawk::brms_SummaryTable(brmsfit, astrology = TRUE) %>%
  mutate(Covariate = str_replace_all(Covariate, "SDM", "SD-"),
         Covariate = str_replace_all(Covariate, "SD-other", "SDMother")) %>% 
  pander(split.cell = 40, split.table = Inf)
} 

Load and clean the data

This section also saves a copy of the cleaned data in the directory data/cleaned_data. If you wish to re-use the data, I suggest you use these cleaned files; the factor and level names are more intuitive than in the original data sheets (however, the actual data are identical).

load_and_clean_data <- function(path){
  
   if(path == "data/messy_data/experiment2.csv"){
   
    expt2 <- read_csv(path) %>%
      mutate(prop_SD_females = female_SD / (female_SD + female_cyo),
             prop_SD_males = male_SD / (male_SD + male_cyo),
             prop_males = (male_SD + male_cyo) / (male_SD + male_cyo + female_SD + female_cyo),
             starting_number_focal_sex = num_embryos,
             SD_offspring = female_SD,
             CyO_offspring = female_cyo) 
    
    expt2$starting_number_focal_sex[expt2$sex_of_embryos == "male"] <- 
      expt2$starting_number_focal_sex[expt2$sex_of_embryos == "male"] -
      expt2$female_SD[expt2$sex_of_embryos == "male"] - expt2$female_cyo[expt2$sex_of_embryos == "male"]
    
    expt2$starting_number_focal_sex[expt2$sex_of_embryos == "female"] <- 
      expt2$starting_number_focal_sex[expt2$sex_of_embryos == "female"] -
      expt2$male_SD[expt2$sex_of_embryos == "female"] - expt2$male_cyo[expt2$sex_of_embryos == "female"]
    
    expt2$SD_offspring[expt2$sex_of_embryos == "male"] <- expt2$male_SD[expt2$sex_of_embryos == "male"]
    expt2$CyO_offspring[expt2$sex_of_embryos == "male"] <- expt2$male_cyo[expt2$sex_of_embryos == "male"]
    
    expt2$starting_number_focal_sex[expt2$starting_number_focal_sex < expt2$SD_focal_sex] <-
      expt2$SD_focal_sex[expt2$starting_number_focal_sex < expt2$SD_focal_sex]
    
    expt2$Dead_offspring <- expt2$starting_number_focal_sex - expt2$SD_offspring - expt2$CyO_offspring
    expt2$Dead_offspring[expt2$Dead_offspring < 0] <- 0
    
    expt2$starting_number_focal_sex <- with(expt2, SD_offspring + CyO_offspring + Dead_offspring)
    
    expt2 <- expt2 %>% 
      select(block, SD_parent, sex_of_embryos, SD, starting_number_focal_sex, SD_offspring, CyO_offspring, Dead_offspring) %>%
      mutate(vial = 1:n(), 
             SD_embryos = "to be estimated",
             CyO_embryos = "to be estimated",
             block = as.numeric(as.factor(block))) %>%
      filter(!(SD %in% c("w1118", "LHm")))
    write_csv(expt2, path = str_replace(path, "messy", "clean")) 
    return(expt2)
  }
  
  # For experiment 1....
  dat <- read_csv(path) %>% 
    mutate(block = as.character(block), 
           copies_of_SD = substr(genotype, 2, 2),
           parent_with_SD = substr(genotype, 1, 1),
           parent_with_SD = replace(parent_with_SD, 
                                    parent_with_SD == "M", "Mother"),
           parent_with_SD = replace(parent_with_SD, 
                                    parent_with_SD == "P", "Father"),
           parent_with_SD = replace(parent_with_SD, 
                                    parent_with_SD == "A", "Both parents"),
           parent_with_SD = replace(parent_with_SD, 
                                    parent_with_SD == "N", "Neither parent"),
           parent_with_SD =  factor(parent_with_SD, 
                                    levels = c("Neither parent", 
                                               "Father", "Mother", 
                                               "Both parents")),
           SD = paste("SD-", SD, sep = ""),
           SD = replace(SD, SD == "SD-N", "No SD chromosome"),
  
           MotherHasSD = ifelse(parent_with_SD != "Father", 1, 0),
           FatherHasSD = ifelse(parent_with_SD != "Mother", 1, 0),
           MotherHasSD = replace(MotherHasSD, SD == "No SD chromosome", 0),
           FatherHasSD = replace(FatherHasSD, SD == "No SD chromosome", 0),
           SD_from_mum = ifelse(MotherHasSD==1 & copies_of_SD=="1", "Yes", "No"),
           SD_from_mum = replace(SD_from_mum, FatherHasSD==1 & MotherHasSD==1 & copies_of_SD=="2", "Yes"),
           SD_from_dad = ifelse(FatherHasSD == 1 & copies_of_SD == "1", "Yes", "No"),
           SD_from_dad = replace(SD_from_dad, FatherHasSD==1 & MotherHasSD==1 & copies_of_SD=="2", "Yes"),
           mum_genotype = relevel(as.factor(ifelse(MotherHasSD==1, SD, "wild_type")), ref = "wild_type"),
           dad_genotype = relevel(as.factor(ifelse(FatherHasSD==1, SD, "wild_type")), ref = "wild_type"))
  
  if(path == "data/messy_data/male_fitness.csv"){
    dat <- dat %>% 
      mutate(focal.male.fitness = focal_daughters + focal_sons,
             rival.male.fitness = rival_daughters + rival_sons,
             total = focal.male.fitness + rival.male.fitness,
             vial = 1:n())
  }
  
  if(path == "data/messy_data/larval_fitness.csv"){
    
    # Note that for 'Mother has SD' crosses (Cross 2), the 'initial_number' of larvae is the maximum possible.
    # The real initial number of non-recombinants is probably a bit lower, since the survival rate of the recombinants is probably not 100%
    # This means the survival rate in Cross 2 is an under-estimate of the true survival rate.
    # For SD-5, it is a very slight underestimate, since SD-5 does not recombine much.
    # For the other two SDs, it is a larger underestimate. 
    
    dat <- dat %>%
      mutate(surviving_females = nonrecombinant_females,
             surviving_males = nonrecombinant_males,
             survivors = nonrecombinant_females + nonrecombinant_males,
             initial_number = number_larvae_in_vial - recombinant_females - recombinant_males,
             number_died = initial_number - survivors,
             larval_density = as.numeric(scale(number_larvae_in_vial)), # mean-centre the larval density covariate
             vial = 1:n(),
             total = survivors + number_died) %>%
      select(-nonrecombinant_females, -nonrecombinant_males, -recombinant_females, -recombinant_males)
      
    # We never put more than 100 larvae in one vial, so for cases where the initial number was 200, it was 2 vials of 100 each, etc.
    dat$larval_density[dat$larval_density > 100] <- 100
    
    # For Cross 1, where both parents have SD, we cannot discriminate larvae with 0 or 1 copies of SD 
    # (until they become adults and develop eyes)
    dat$copies_of_SD[dat$genotype == "A0+A1"] <- "0 or 1"
  }
  
  if(path == "data/messy_data/female_fitness.csv"){
    dat$vial <- 1:nrow(dat)
  }
  
  # 'genotype' variable is a composite of SD genotype and parental origin, used during data collection. Not needed for analysis
  dat <- dat %>% select(-genotype) 
  
  # save the clean data
  write_csv(dat, path = str_replace(path, "messy", "clean")) 
  dat 
}

larvae  <- load_and_clean_data("data/messy_data/larval_fitness.csv") 
males   <- load_and_clean_data("data/messy_data/male_fitness.csv") 
females <- load_and_clean_data("data/messy_data/female_fitness.csv") 
expt2   <- load_and_clean_data("data/messy_data/experiment2.csv") 

Summary statistics for the four response variables

Larval survival data

Table S1: Number and percentage of L1 larvae surviving to adulthood for each SD genotype and cross type.

tidy_names <- function(df){
  df <- as.data.frame(df)
  names(df) <- str_replace_all(names(df), "_", " ")
  names(df) <- str_replace_all(names(df), "[.]", " ")
  names(df) <- sapply(names(df), function(x) {
    if(substr(x,1,2) == "n ") return(x)
    substr(x,1,1) <- toupper(substr(x,1,1))
    x
  })
  names(df) <- str_replace_all(names(df), "[Pp]ercent", "%")
  names(df) <- str_replace_all(names(df), "Number of", "n")
  df
}


tab_S1 <- larvae %>% 
  group_by(SD, copies_of_SD, parent_with_SD) %>% 
  summarise(number_larvae_counted = sum(survivors + number_died),
            number_of_survivors = sum(survivors),
            percent_surviving = 100 * (number_of_survivors / number_larvae_counted)) %>%
    mutate(percent_surviving = format(round(percent_surviving, 1), nsmall = 1)) %>%
  tidy_names() 
saveRDS(tab_S1, file = "output/tab_S1.rds")
tab_S1 %>% pander(split.cell = 40, split.table = Inf)
SD Copies of SD Parent with SD Number larvae counted n survivors % surviving
No SD chromosome 0 Neither parent 600 495 82.5
SD-5 0 Father 113 89 78.8
SD-5 0 Mother 459 408 88.9
SD-5 0 or 1 Both parents 700 520 74.3
SD-5 1 Father 563 412 73.2
SD-5 1 Mother 494 415 84.0
SD-5 2 Both parents 40 0 0.0
SD-72 0 Father 287 226 78.7
SD-72 0 Mother 396 333 84.1
SD-72 0 or 1 Both parents 700 542 77.4
SD-72 1 Father 600 477 79.5
SD-72 1 Mother 423 342 80.9
SD-72 2 Both parents 600 0 0.0
SD-Mad 0 Father 296 239 80.7
SD-Mad 0 Mother 371 279 75.2
SD-Mad 0 or 1 Both parents 700 558 79.7
SD-Mad 1 Father 600 462 77.0
SD-Mad 1 Mother 436 320 73.4
SD-Mad 2 Both parents 585 413 70.6

Adult sex ratio

Table S2: Number and percentage of male and female adults emerging from the juvenile fitness assay vials.

tab_S2 <- larvae %>% 
  group_by(SD, copies_of_SD, parent_with_SD) %>% 
  summarise(n.males = sum(surviving_males),
            n.females = sum(surviving_females),
            n.total = sum(survivors), 
            percent.male = format(round(100 * sum(surviving_males) / n.total, 1), nsmall = 1)) %>%
  tidy_names() 

saveRDS(tab_S2, file = "output/tab_S2.rds")
tab_S2 %>% pander(split.cell = 40, split.table = Inf)
SD Copies of SD Parent with SD n males n females n total % male
No SD chromosome 0 Neither parent 267 228 495 53.9
SD-5 0 Father 48 41 89 53.9
SD-5 0 Mother 206 202 408 50.5
SD-5 0 or 1 Both parents 239 281 520 46.0
SD-5 1 Father 196 216 412 47.6
SD-5 1 Mother 193 222 415 46.5
SD-5 2 Both parents 0 0 0 NaN
SD-72 0 Father 105 121 226 46.5
SD-72 0 Mother 169 164 333 50.8
SD-72 0 or 1 Both parents 272 270 542 50.2
SD-72 1 Father 233 244 477 48.8
SD-72 1 Mother 186 156 342 54.4
SD-72 2 Both parents 0 0 0 NaN
SD-Mad 0 Father 102 137 239 42.7
SD-Mad 0 Mother 145 134 279 52.0
SD-Mad 0 or 1 Both parents 253 305 558 45.3
SD-Mad 1 Father 190 272 462 41.1
SD-Mad 1 Mother 184 136 320 57.5
SD-Mad 2 Both parents 209 204 413 50.6

Male fitness data

Table S4: Average relative fitness of adult males for each SD genotype and cross type, expressed as the average proportion of offspring sired. The last two columns give the sample size in terms of number of vials (each of which contained 5 focal males), and number of males.

SE <- function(x) sd(x) / sqrt(length(x))
tab_S4 <- males %>% 
  group_by(SD, copies_of_SD, parent_with_SD) %>% 
  summarise(average_relative_fitness = mean(focal.male.fitness / (focal.male.fitness + rival.male.fitness)),
            SE = SE(focal.male.fitness / (focal.male.fitness + rival.male.fitness)),
            number_of_vials = n(),
            number_of_males = number_of_vials * 5) %>%
  mutate(average_relative_fitness = format(round(average_relative_fitness, 2), nsmall = 2),
         SE = format(round(SE, 3), nsmall = 3)) %>%
  tidy_names()

saveRDS(tab_S4, file = "output/tab_S4.rds")
tab_S4 %>% pander(split.cell = 40, split.table = Inf)
SD Copies of SD Parent with SD Average relative fitness SE n vials n males
No SD chromosome 0 Neither parent 0.79 0.040 13 65
SD-5 0 Father 0.68 0.147 5 25
SD-5 0 Mother 0.82 0.045 17 85
SD-5 0 Both parents 0.59 NA 1 5
SD-5 1 Father 0.14 0.055 18 90
SD-5 1 Mother 0.32 0.072 12 60
SD-5 1 Both parents 0.39 0.074 13 65
SD-72 0 Father 0.88 0.027 16 80
SD-72 0 Mother 0.77 0.054 13 65
SD-72 0 Both parents 0.79 NA 1 5
SD-72 1 Father 0.76 0.045 18 90
SD-72 1 Mother 0.80 0.039 17 85
SD-72 1 Both parents 0.67 0.051 19 95
SD-Mad 0 Father 0.75 0.055 14 70
SD-Mad 0 Mother 0.75 0.069 11 55
SD-Mad 0 Both parents 0.82 0.078 5 25
SD-Mad 1 Father 0.87 0.028 18 90
SD-Mad 1 Mother 0.81 0.037 17 85
SD-Mad 1 Both parents 0.75 0.053 18 90
SD-Mad 2 Both parents 0.19 0.072 14 70

Female fitness data

Table S3: Average fecundity of adult females for each SD genotype and cross type. The last two columns give the sample size in terms of number of oviposition vials (each of which contained up to 5 focal females), and number of males.

tab_S3 <- females %>% 
  mutate(larvae_per_female = larvae_produced / number_of_laying_females) %>%
  group_by(SD, copies_of_SD, parent_with_SD) %>% 
  summarise(average_fecundity = mean(larvae_per_female),
            SE = SE(larvae_per_female),
            number_of_vials = n(), 
            number_of_females = sum(number_of_laying_females)) %>%
  mutate(average_fecundity = format(round(average_fecundity, 2), nsmall = 2),
         SE = format(round(SE, 3), nsmall = 3)) %>%
  tidy_names() 

saveRDS(tab_S3, file = "output/tab_S3.rds")
tab_S3 %>% pander(split.cell = 40, split.table = Inf)
SD Copies of SD Parent with SD Average fecundity SE n vials n females
No SD chromosome 0 Neither parent 26.55 3.874 10 48
SD-5 0 Father 25.06 8.127 6 28
SD-5 0 Mother 41.13 3.700 15 71
SD-5 0 Both parents 24.95 3.767 5 22
SD-5 1 Father 28.88 3.337 12 55
SD-5 1 Mother 29.74 3.118 16 69
SD-5 1 Both parents 26.83 2.583 15 67
SD-72 0 Father 32.68 3.258 14 65
SD-72 0 Mother 35.10 3.023 15 68
SD-72 0 Both parents 22.53 6.671 3 15
SD-72 1 Father 33.97 2.743 16 77
SD-72 1 Mother 41.90 3.792 14 68
SD-72 1 Both parents 31.85 2.885 15 73
SD-Mad 0 Father 28.25 3.769 16 79
SD-Mad 0 Mother 44.71 3.723 13 65
SD-Mad 0 Both parents 16.50 1.762 3 14
SD-Mad 1 Father 36.85 3.968 16 77
SD-Mad 1 Mother 40.58 4.602 14 64
SD-Mad 1 Both parents 34.88 3.478 16 76
SD-Mad 2 Both parents 11.26 1.631 17 83

Experiment 2 larval survival data

Table S5: Number and percentage of L1 larvae surviving to adulthood in Experiment 2, for each SD genotype, cross type, and offspring sex.

tab_S5 <- expt2 %>%
  rename(parent_with_SD = SD_parent, offspring_sex = sex_of_embryos) %>%
  mutate(SD_embryos = SD_offspring + rbinom(n(), Dead_offspring, 0.5),
         CyO_embryos = starting_number_focal_sex - SD_embryos,
         parent_with_SD = ifelse(parent_with_SD == "dad", "Father", "Mother"),
         offspring_sex = ifelse(offspring_sex == "female", "Female", "Male")) %>%
  group_by(SD, parent_with_SD, offspring_sex) %>% 
  summarise(percent_surviving_SD_larvae = mean(100 * SD_offspring / SD_embryos),
            percent_surviving_CyO_larvae = mean(100 * CyO_offspring / CyO_embryos),
            n_larvae_counted = sum(starting_number_focal_sex),
            n_crosses = n()) %>%
  mutate(percent_surviving_SD_larvae = format(round(percent_surviving_SD_larvae, 1), nsmall = 1),
         percent_surviving_CyO_larvae = format(round(percent_surviving_CyO_larvae, 1), nsmall = 1)) %>%
  tidy_names()

saveRDS(tab_S5, file = "output/tab_S5.rds")
tab_S5 %>% pander(split.cell = 40, split.table = Inf)
SD Parent with SD Offspring sex % surviving SD larvae % surviving CyO larvae n larvae counted n crosses
SD-5 Father Female 83.4 83.8 763 16
SD-5 Father Male 79.0 79.3 727 17
SD-5 Mother Female 76.4 81.9 871 18
SD-5 Mother Male 70.8 67.8 972 20
SD-72 Father Female 85.5 85.9 744 16
SD-72 Father Male 81.0 78.6 615 15
SD-72 Mother Female 90.2 87.4 1123 23
SD-72 Mother Male 78.9 76.8 1186 24
SD-Mad Father Female 87.2 83.9 457 10
SD-Mad Father Male 87.8 83.1 480 11
SD-Mad Mother Female 84.4 85.7 942 20
SD-Mad Mother Male 82.3 78.4 1010 21

Analysis of L1 larva-to-adult survival

Run a Bayesian binomial generalised linear mixed model

if(!file.exists("data/model_output/larvae_brms.rds")){
  
  larvae_brms <-  brm(survivors | trials(total) ~ larval_density + SD * copies_of_SD * parent_with_SD + (1 | vial), 
                      family = binomial,
                      data = larvae, 
                      iter = 4000, chains = 4, cores = nCores,
                      control = list(adapt_delta = 0.99, max_treedepth = 15),
                      prior = prior(normal(0, 5), class = "b"))
  
  saveRDS(larvae_brms, "data/model_output/larvae_brms.rds")
} else larvae_brms <- readRDS("data/model_output/larvae_brms.rds")

Check a diagnostic plot of the model

This plot shows the frequency distribution of the original data (black line), along with predicted data that were computed using 10 random samples from the posterior of the model parameters. The posterior predicted values closely match the originals, suggesting that the model is approximating the data reasonably well.

pp_check(larvae_brms)

Version Author Date
437b6d7 lukeholman 2019-06-11
8d92589 lukeholman 2019-03-12
6b03f9c lukeholman 2019-03-12

Inspect the model summary

This summary shows the median, error, and 95% CIs for the posterior for all the fixed effects. This table is difficult to interpret and not very informative, but is presented here for completeness.

summarise_brms(larvae_brms)
Covariate Estimate Est.Error l-95% CI u-95% CI Notable
Intercept 1.64 0.34 0.97 2.32 *
larval_density -0.04 0.17 -0.37 0.31
SDSD-5 -0.57 3.18 -6.69 5.57
SDSD-72 -0.79 3.18 -7.07 5.43
SDSD-Mad 0.19 3.17 -6.16 6.49
copies_of_SD0or1 0.89 4.29 -7.55 9.18
copies_of_SD1 -0.07 3.55 -7.08 6.88
copies_of_SD2 -2.56 4.31 -11.00 5.91
parent_with_SDFather 0.04 3.06 -5.90 5.98
parent_with_SDMother 0.35 3.04 -5.64 6.33
parent_with_SDBothparents -1.75 4.10 -9.63 6.33
SDSD-5:copies_of_SD0or1 0.78 4.29 -7.69 9.14
SDSD-72:copies_of_SD0or1 1.04 4.40 -7.63 9.81
SDSD-Mad:copies_of_SD0or1 -0.89 4.31 -9.41 7.68
SDSD-5:copies_of_SD1 -0.16 3.55 -7.05 6.74
SDSD-72:copies_of_SD1 0.08 3.53 -6.79 6.97
SDSD-Mad:copies_of_SD1 -0.03 3.54 -6.97 6.90
SDSD-5:copies_of_SD2 -2.16 4.53 -11.22 6.71
SDSD-72:copies_of_SD2 -2.78 4.65 -11.86 6.34
SDSD-Mad:copies_of_SD2 2.35 4.41 -6.37 11.11
SDSD-5:parent_with_SDFather 0.03 3.42 -6.59 6.91
SDSD-72:parent_with_SDFather 0.37 3.46 -6.53 7.19
SDSD-Mad:parent_with_SDFather -0.45 3.45 -7.31 6.26
SDSD-5:parent_with_SDMother 0.73 3.42 -5.87 7.58
SDSD-72:parent_with_SDMother 0.51 3.32 -6.05 7.12
SDSD-Mad:parent_with_SDMother -0.89 3.37 -7.41 5.86
SDSD-5:parent_with_SDBothparents -1.38 4.16 -9.58 6.70
SDSD-72:parent_with_SDBothparents -1.65 4.26 -9.96 6.89
SDSD-Mad:parent_with_SDBothparents 1.37 4.13 -6.77 9.69
copies_of_SD0or1:parent_with_SDFather -0.01 5.11 -10.16 10.02
copies_of_SD1:parent_with_SDFather 0.03 3.54 -6.90 6.94
copies_of_SD2:parent_with_SDFather 0.11 4.99 -9.60 9.91
copies_of_SD0or1:parent_with_SDMother -0.04 4.91 -9.60 9.63
copies_of_SD1:parent_with_SDMother -0.13 3.63 -7.24 7.09
copies_of_SD2:parent_with_SDMother 0.06 5.06 -9.76 9.81
copies_of_SD0or1:parent_with_SDBothparents 0.89 4.23 -7.16 9.22
copies_of_SD1:parent_with_SDBothparents 0.08 4.90 -9.61 9.82
copies_of_SD2:parent_with_SDBothparents -2.68 4.37 -11.18 5.82
SDSD-5:copies_of_SD0or1:parent_with_SDFather 0.01 4.93 -9.76 9.63
SDSD-72:copies_of_SD0or1:parent_with_SDFather -0.03 5.09 -10.10 10.02
SDSD-Mad:copies_of_SD0or1:parent_with_SDFather 0.00 5.08 -9.93 9.91
SDSD-5:copies_of_SD1:parent_with_SDFather 0.06 3.53 -6.84 6.90
SDSD-72:copies_of_SD1:parent_with_SDFather 0.11 3.55 -6.86 7.07
SDSD-Mad:copies_of_SD1:parent_with_SDFather -0.11 3.57 -7.17 6.86
SDSD-5:copies_of_SD2:parent_with_SDFather 0.09 5.06 -9.60 9.95
SDSD-72:copies_of_SD2:parent_with_SDFather 0.04 4.94 -9.58 9.60
SDSD-Mad:copies_of_SD2:parent_with_SDFather -0.01 4.96 -9.76 9.88
SDSD-5:copies_of_SD0or1:parent_with_SDMother 0.03 5.09 -9.86 10.08
SDSD-72:copies_of_SD0or1:parent_with_SDMother 0.02 4.96 -9.67 9.71
SDSD-Mad:copies_of_SD0or1:parent_with_SDMother -0.03 4.97 -9.89 9.64
SDSD-5:copies_of_SD1:parent_with_SDMother -0.14 3.60 -7.19 6.93
SDSD-72:copies_of_SD1:parent_with_SDMother -0.10 3.64 -7.21 7.11
SDSD-Mad:copies_of_SD1:parent_with_SDMother 0.12 3.58 -6.92 7.18
SDSD-5:copies_of_SD2:parent_with_SDMother 0.01 4.86 -9.39 9.50
SDSD-72:copies_of_SD2:parent_with_SDMother -0.02 5.06 -9.93 9.87
SDSD-Mad:copies_of_SD2:parent_with_SDMother 0.04 5.02 -9.64 9.90
SDSD-5:copies_of_SD0or1:parent_with_SDBothparents 0.76 4.26 -7.48 8.94
SDSD-72:copies_of_SD0or1:parent_with_SDBothparents 0.99 4.41 -7.54 9.71
SDSD-Mad:copies_of_SD0or1:parent_with_SDBothparents -0.85 4.30 -9.30 7.38
SDSD-5:copies_of_SD1:parent_with_SDBothparents -0.05 4.89 -9.48 9.44
SDSD-72:copies_of_SD1:parent_with_SDBothparents 0.09 5.12 -9.96 10.17
SDSD-Mad:copies_of_SD1:parent_with_SDBothparents 0.00 4.97 -9.69 9.78
SDSD-5:copies_of_SD2:parent_with_SDBothparents -2.17 4.58 -11.16 6.77
SDSD-72:copies_of_SD2:parent_with_SDBothparents -2.80 4.54 -11.78 5.99
SDSD-Mad:copies_of_SD2:parent_with_SDBothparents 2.32 4.36 -6.35 10.75

Hypothesis testing

Table S6: The results of hypothesis tests computed using the model of larval survival in Experiment 1. Each row gives the posterior estimate of a difference in means, such that the estimate is positive if mean 1 is larger than mean 2, and negative otherwise (expressed in % larval survival). The mean 1 and mean 2 columns list the parent which had SD (mother, father, or both), followed by the number of SD alleles present in the offspring (0, 1 or 2). The Posterior probability column gives the probability that the mean with the smaller point estimate is actually larger than the other mean, analagously to a one-tailed p-value. Asterisks highlight rows where the posterior probability is less than 0.05.

# Function to compute posterior difference in means, with the posterior_probability of the smaller mean actually being the larger one (plus the larger evidence ratio, e.g. ER = 10 means it's 10x more likely the smaller mean really is smaller not larger, ER=1 means it's equally likely the smaller one is actually larger)
compare <- function(mean1, mean2){
  difference <- mean2 - mean1
  prob_less <- hypothesis(data.frame(x = difference), "x < 0")$hypothesis
  prob_more <- hypothesis(data.frame(x = difference), "x > 0")$hypothesis
  hypothesis(data.frame(x = difference), "x = 0")$hypothesis %>% 
    mutate(Posterior_probability = min(c(prob_less$Post.Prob, prob_more$Post.Prob)),
           Evidence_ratio = max(c(prob_less$Evid.Ratio, prob_more$Evid.Ratio)))
}

hypothesis_tests <- function(model, dat, SR = FALSE, divisor = NULL){
  
  SDs <- c("SD-5", "SD-72", "SD-Mad")
  
  new <- dat %>%
    select(SD, copies_of_SD, MotherHasSD, FatherHasSD, SD_from_mum, 
           SD_from_dad, mum_genotype, dad_genotype, parent_with_SD) %>%
    distinct() %>%
    arrange(SD, copies_of_SD, MotherHasSD, FatherHasSD, SD_from_mum, 
            SD_from_dad, mum_genotype, dad_genotype) %>%
    mutate(total = 100, 
           number_of_laying_females = 5, 
           larval_density = 0,
           copies_of_SD = as.character(copies_of_SD),
           i = 1:n(),
           parent = ifelse(FatherHasSD == 1, "B", "M"),
           parent = replace(parent, MotherHasSD == 0, "F"))
  
  
  SD5 <- new %>% filter(SD == "SD-5")
  SD72 <- new %>% filter(SD == "SD-72")
  SDMad <- new %>% filter(SD == "SD-Mad")
  
  pred <- fitted(model, re_formula = NA,  
                 newdata = new, 
                 summary = FALSE) 
  
  if(!is.null(divisor)) pred <- pred / divisor
  
  control_0SD_neither <- pred[, new %>% filter(SD == "No SD chromosome") %>% pull(i)]
  
  SD5_0SD_mother <- pred[, SD5 %>% filter(copies_of_SD == "0" & parent == "M") %>% pull(i)]
  SD5_1SD_mother <- pred[, SD5 %>% filter(copies_of_SD == "1" & parent == "M") %>% pull(i)]
  SD5_0SD_father <- pred[, SD5 %>% filter(copies_of_SD == "0" & parent == "F") %>% pull(i)]
  SD5_1SD_father <- pred[, SD5 %>% filter(copies_of_SD == "1" & parent == "F") %>% pull(i)]
  
  SD72_0SD_mother <- pred[, SD72 %>% filter(copies_of_SD == "0" & parent == "M") %>% pull(i)]
  SD72_1SD_mother <- pred[, SD72 %>% filter(copies_of_SD == "1" & parent == "M") %>% pull(i)]
  SD72_0SD_father <- pred[, SD72 %>% filter(copies_of_SD == "0" & parent == "F") %>% pull(i)]
  SD72_1SD_father <- pred[, SD72 %>% filter(copies_of_SD == "1" & parent == "F") %>% pull(i)]
  
  SDMad_0SD_mother <- pred[, SDMad %>% filter(copies_of_SD == "0" & parent == "M") %>% pull(i)]
  SDMad_1SD_mother <- pred[, SDMad %>% filter(copies_of_SD == "1" & parent == "M") %>% pull(i)]
  SDMad_0SD_father <- pred[, SDMad %>% filter(copies_of_SD == "0" & parent == "F") %>% pull(i)]
  SDMad_1SD_father <- pred[, SDMad %>% filter(copies_of_SD == "1" & parent == "F") %>% pull(i)]
  
  SDMad_1SD_both <- pred[, SDMad %>% filter(copies_of_SD %in% c("0 or 1", "1") & parent == "B") %>% pull(i)]
  SDMad_2 <- pred[, SDMad %>% filter(copies_of_SD == "2") %>% pull(i)]
  
  # For larvae data only, we also have observations of SD-5 and SD-Mad homozygotes
  if(sum(new$copies_of_SD == "2") > 1){ 
    SD5_1SD_both  <- pred[, SD5  %>% filter(copies_of_SD %in% c("0 or 1", "1") & parent == "B") %>% pull(i)]
    SD72_1SD_both <- pred[, SD72 %>% filter(copies_of_SD %in% c("0 or 1", "1") & parent == "B") %>% pull(i)]
    SD5_2  <- pred[, SD5  %>% filter(copies_of_SD == "2") %>% pull(i)]
    SD72_2 <- pred[, SD72 %>% filter(copies_of_SD == "2") %>% pull(i)]
    
    two_compare <- data.frame(
      SD = SDs,
      Mean_1 = rep(paste("Both parents,", ifelse("0 or 1" %in% SD5$copies_of_SD, "0 or 1", "1")), 3),
      Mean_2 = rep("Both parents, 2", 3),
      rbind(
        compare(SD5_1SD_both, SD5_2),
        compare(SD72_1SD_both, SD72_2),
        compare(SDMad_1SD_both, SDMad_2)
      ) 
    )
    
  } else { # for non-larva data
    two_compare <- data.frame(SD = "SD-Mad", 
                              Mean_1 = "Both parents, 1",
                              Mean_2 = "Both parents, 2", 
                              compare(SDMad_1SD_both, SDMad_2)) 
  }
  
  if("0 or 1" %in% new$copies_of_SD){  
    SD5_0or1   <- pred[, SD5 %>% filter(copies_of_SD == "0 or 1") %>% pull(i)]
    SD72_0or1  <- pred[, SD72 %>% filter(copies_of_SD == "0 or 1") %>% pull(i)]
    SDMad_0or1 <- pred[, SDMad %>% filter(copies_of_SD == "0 or 1") %>% pull(i)]
  }
  
  rbind(
    data.frame(
      SD = rep(SDs, 6),
      Mean_1 = c(rep("Neither, 0", 6), rep("Mother, 0", 3), rep("Mother, 1", 3), rep("Mother, 0", 3), rep("Father, 0", 3)),
      Mean_2 = c(rep("Father, 0", 3), rep("Mother, 0", 3),  
                 rep("Father, 0", 3), rep("Father, 1", 3), rep("Mother, 1", 3), rep("Father, 1", 3)),
      rbind(
        
        compare(control_0SD_neither, SD5_0SD_father), # controls vs 0 SD individuals
        compare(control_0SD_neither, SD72_0SD_father),
        compare(control_0SD_neither, SDMad_0SD_father),
        compare(control_0SD_neither, SD5_0SD_mother),
        compare(control_0SD_neither, SD72_0SD_mother),
        compare(control_0SD_neither, SDMad_0SD_mother),
        
        compare(SD5_0SD_mother, SD5_0SD_father), # parental effects on 0 SD individuals
        compare(SD72_0SD_mother, SD72_0SD_father),
        compare(SDMad_0SD_mother, SDMad_0SD_father),
        
        compare(SD5_1SD_mother, SD5_1SD_father), # parental effects on 1 SD individuals
        compare(SD72_1SD_mother, SD72_1SD_father),
        compare(SDMad_1SD_mother, SDMad_1SD_father),
        
        compare(SD5_0SD_mother, SD5_1SD_mother), # effect of inheriting SD, from the mother
        compare(SD72_0SD_mother, SD72_1SD_mother),
        compare(SDMad_0SD_mother, SDMad_1SD_mother),
        
        compare(SD5_0SD_father, SD5_1SD_father), # effect of inheriting SD, from the father
        compare(SD72_0SD_father, SD72_1SD_father),
        compare(SDMad_0SD_father, SDMad_1SD_father)
        
      )), 
    two_compare) %>% 
    select(-Hypothesis, -Evid.Ratio, -Post.Prob, -Star) %>%
    mutate(Notable = ifelse(Posterior_probability < 0.05, "*", "")) 
}

clean_table <- function(tabl){
  cols <- c("Trait", "SD", "Comparison", "Difference", "Error", "Posterior_probability")
  if(!("Trait" %in% names(tabl))){
    cols <- cols[cols != "Trait"]
  }
  
  tabl <- mutate(tabl, 
                 Estimate = format(round(Estimate, 1), nsmall = 1),  
                 Est.Error = format(round(Est.Error, 1), nsmall = 1),
                 CI.Lower = format(round(CI.Lower, 1), nsmall = 1),
                 CI.Upper = format(round(CI.Upper, 1), nsmall = 1), 
                 Evidence_ratio = format(round(Evidence_ratio, 1), nsmall = 1), 
                 Difference = paste(Estimate, " (", CI.Lower, " to ", CI.Upper, ")", sep = ""),
                 Comparison = paste(Mean_1, "-", Mean_2),
                 Error = Est.Error) %>%
    select(!! cols)
  tabl %>% tidy_names()
}

tests1 <- hypothesis_tests(larvae_brms, larvae)
tab_S6 <- tests1 %>% clean_table() 
saveRDS(tab_S6, file = "output/tab_S6.rds")
tab_S6 %>% pander(split.cell = 40, split.table = Inf)
SD Comparison Difference Error Posterior probability
SD-5 Neither, 0 - Father, 0 -8.9 (-34.9 to 11.6) 11.8 0.2254
SD-72 Neither, 0 - Father, 0 -6.0 (-24.3 to 10.5) 8.9 0.2464
SD-Mad Neither, 0 - Father, 0 -3.5 (-21.0 to 12.5) 8.4 0.3415
SD-5 Neither, 0 - Mother, 0 5.8 ( -5.3 to 18.1) 5.9 0.1557
SD-72 Neither, 0 - Mother, 0 0.9 (-12.1 to 14.0) 6.6 0.4435
SD-Mad Neither, 0 - Mother, 0 -5.4 (-20.0 to 9.3) 7.3 0.2181
SD-5 Mother, 0 - Father, 0 -14.7 (-39.3 to 2.8) 10.7 0.0575
SD-72 Mother, 0 - Father, 0 -6.9 (-25.0 to 9.6) 8.8 0.2145
SD-Mad Mother, 0 - Father, 0 1.9 (-15.9 to 18.1) 8.6 0.3991
SD-5 Mother, 1 - Father, 1 -10.6 (-26.4 to 4.4) 7.8 0.07975
SD-72 Mother, 1 - Father, 1 -1.2 (-15.4 to 13.1) 7.2 0.429
SD-Mad Mother, 1 - Father, 1 1.0 (-15.2 to 17.9) 8.4 0.4487
SD-5 Mother, 0 - Mother, 1 -5.6 (-17.4 to 5.2) 5.7 0.156
SD-72 Mother, 0 - Mother, 1 -3.0 (-16.9 to 10.2) 6.8 0.326
SD-Mad Mother, 0 - Mother, 1 -1.9 (-19.4 to 14.4) 8.6 0.4131
SD-5 Father, 0 - Father, 1 -1.6 (-23.4 to 25.5) 12.5 0.4146
SD-72 Father, 0 - Father, 1 2.7 (-15.1 to 21.5) 9.3 0.3935
SD-Mad Father, 0 - Father, 1 -2.8 (-20.4 to 16.7) 9.2 0.3656
SD-5 Both parents, 0 or 1 - Both parents, 2 -77.1 (-87.8 to -62.2) 6.5 0
SD-72 Both parents, 0 or 1 - Both parents, 2 -77.5 (-87.8 to -63.6) 6.1 0
SD-Mad Both parents, 0 or 1 - Both parents, 2 -10.6 (-27.7 to 6.7) 8.7 0.1046

Analysis of adult sex ratio

Run a Bayesian binomial generalised linear mixed model

SR_data <- larvae %>% filter(survivors > 0) %>% 
  mutate(Males = surviving_males, 
         total = surviving_males + surviving_females)

if(!file.exists("data/model_output/SR_brms.rds")){
  SR_brms <-  brm(Males | trials(total) ~ SD * copies_of_SD * parent_with_SD + (1 | vial), 
                  family = binomial,
                  data = SR_data, iter = 4000, chains = 4, cores = nCores,
                  control = list(adapt_delta = 0.99, max_treedepth = 15), 
                  prior = prior(normal(0, 5), class = "b"))
  
  saveRDS(SR_brms, "data/model_output/SR_brms.rds")
} else SR_brms <- readRDS("data/model_output/SR_brms.rds")

Check a diagnostic plot of the model

This plot shows the frequency distribution of the original data (black line), along with predicted data that were computed using 10 random samples from the posterior of the model parameters. The posterior predicted values closely match the originals, suggesting that the model is approximating the data reasonably well.

pp_check(SR_brms)

Version Author Date
437b6d7 lukeholman 2019-06-11
8d92589 lukeholman 2019-03-12
6b03f9c lukeholman 2019-03-12

Inspect the model summary

This summary shows the median, error, and 95% CIs for the posterior for all the fixed effects. This table is difficult to interpret and not very informative, but is presented here for completeness.

summarise_brms(SR_brms)
Covariate Estimate Est.Error l-95% CI u-95% CI Notable
Intercept 0.15 0.16 -0.16 0.46
SDSD-5 0.06 3.20 -6.28 6.28
SDSD-72 -0.08 3.19 -6.29 6.31
SDSD-Mad -0.13 3.15 -6.34 5.91
copies_of_SD0or1 -0.10 4.36 -8.53 8.54
copies_of_SD1 0.03 3.55 -6.99 6.99
copies_of_SD2 -0.01 4.54 -9.04 8.98
parent_with_SDFather -0.12 3.05 -6.15 5.89
parent_with_SDMother -0.01 3.07 -5.99 5.92
parent_with_SDBothparents -0.03 4.15 -8.19 8.13
SDSD-5:copies_of_SD0or1 -0.11 4.31 -8.51 8.49
SDSD-72:copies_of_SD0or1 0.04 4.35 -8.57 8.45
SDSD-Mad:copies_of_SD0or1 -0.07 4.27 -8.43 8.10
SDSD-5:copies_of_SD1 -0.17 3.61 -7.23 6.96
SDSD-72:copies_of_SD1 0.05 3.53 -6.84 6.96
SDSD-Mad:copies_of_SD1 0.03 3.53 -6.85 6.93
SDSD-5:copies_of_SD2 0.01 5.00 -9.76 9.78
SDSD-72:copies_of_SD2 0.02 5.02 -9.82 9.99
SDSD-Mad:copies_of_SD2 0.01 4.47 -8.64 8.77
SDSD-5:parent_with_SDFather 0.07 3.42 -6.61 6.85
SDSD-72:parent_with_SDFather -0.08 3.37 -6.74 6.44
SDSD-Mad:parent_with_SDFather -0.20 3.41 -6.80 6.50
SDSD-5:parent_with_SDMother -0.15 3.48 -6.98 6.78
SDSD-72:parent_with_SDMother -0.03 3.41 -6.73 6.52
SDSD-Mad:parent_with_SDMother 0.11 3.42 -6.61 6.87
SDSD-5:parent_with_SDBothparents -0.03 4.31 -8.53 8.36
SDSD-72:parent_with_SDBothparents -0.01 4.40 -8.72 8.59
SDSD-Mad:parent_with_SDBothparents -0.01 4.17 -8.43 8.09
copies_of_SD0or1:parent_with_SDFather -0.01 4.91 -9.76 9.72
copies_of_SD1:parent_with_SDFather -0.02 3.50 -7.03 6.85
copies_of_SD2:parent_with_SDFather 0.06 4.97 -9.73 9.93
copies_of_SD0or1:parent_with_SDMother 0.02 5.00 -9.91 9.73
copies_of_SD1:parent_with_SDMother 0.06 3.49 -6.85 6.98
copies_of_SD2:parent_with_SDMother 0.01 5.08 -9.92 10.00
copies_of_SD0or1:parent_with_SDBothparents -0.05 4.32 -8.58 8.30
copies_of_SD1:parent_with_SDBothparents 0.03 4.98 -9.69 9.86
copies_of_SD2:parent_with_SDBothparents 0.05 4.48 -8.89 8.87
SDSD-5:copies_of_SD0or1:parent_with_SDFather -0.04 4.95 -9.71 9.82
SDSD-72:copies_of_SD0or1:parent_with_SDFather 0.03 5.11 -9.90 10.00
SDSD-Mad:copies_of_SD0or1:parent_with_SDFather 0.03 4.98 -9.82 9.78
SDSD-5:copies_of_SD1:parent_with_SDFather -0.09 3.50 -6.93 6.85
SDSD-72:copies_of_SD1:parent_with_SDFather 0.04 3.54 -6.93 6.90
SDSD-Mad:copies_of_SD1:parent_with_SDFather -0.09 3.54 -7.00 6.89
SDSD-5:copies_of_SD2:parent_with_SDFather 0.04 5.01 -9.56 9.79
SDSD-72:copies_of_SD2:parent_with_SDFather 0.00 4.84 -9.52 9.55
SDSD-Mad:copies_of_SD2:parent_with_SDFather -0.04 5.05 -9.75 10.02
SDSD-5:copies_of_SD0or1:parent_with_SDMother -0.09 5.05 -10.00 9.75
SDSD-72:copies_of_SD0or1:parent_with_SDMother -0.01 4.92 -9.78 9.66
SDSD-Mad:copies_of_SD0or1:parent_with_SDMother 0.01 4.97 -9.72 9.53
SDSD-5:copies_of_SD1:parent_with_SDMother -0.12 3.57 -7.11 6.90
SDSD-72:copies_of_SD1:parent_with_SDMother 0.02 3.44 -6.68 6.75
SDSD-Mad:copies_of_SD1:parent_with_SDMother 0.16 3.50 -6.79 7.13
SDSD-5:copies_of_SD2:parent_with_SDMother 0.05 4.94 -9.42 9.67
SDSD-72:copies_of_SD2:parent_with_SDMother -0.02 5.01 -9.83 9.61
SDSD-Mad:copies_of_SD2:parent_with_SDMother 0.02 4.95 -9.76 9.99
SDSD-5:copies_of_SD0or1:parent_with_SDBothparents -0.05 4.37 -8.45 8.44
SDSD-72:copies_of_SD0or1:parent_with_SDBothparents 0.08 4.45 -8.39 8.82
SDSD-Mad:copies_of_SD0or1:parent_with_SDBothparents 0.05 4.25 -8.36 8.33
SDSD-5:copies_of_SD1:parent_with_SDBothparents 0.05 4.97 -9.73 9.99
SDSD-72:copies_of_SD1:parent_with_SDBothparents -0.03 5.04 -10.04 9.77
SDSD-Mad:copies_of_SD1:parent_with_SDBothparents 0.01 4.95 -9.71 9.58
SDSD-5:copies_of_SD2:parent_with_SDBothparents 0.00 5.06 -9.77 10.07
SDSD-72:copies_of_SD2:parent_with_SDBothparents -0.04 4.93 -9.73 9.63
SDSD-Mad:copies_of_SD2:parent_with_SDBothparents -0.01 4.54 -8.94 8.88

Hypothesis testing

Table S7: The results of hypothesis tests computed using the model of adult sex ratio in Experiment 1. Each row gives the posterior estimate of a difference in means, such that the estimate is positive if mean 1 is larger than mean 2, and negative otherwise (expressed in % males). The mean 1 and mean 2 columns list the parent which had SD (mother, father, or both), followed by the number of SD alleles present in the offspring (0, 1 or 2). The Posterior probability column gives the probability that the mean with the smaller point estimate is actually larger than the other mean, analagously to a one-tailed p-value. Asterisks highlight rows where the posterior probability is less than 0.05.

tests2 <- hypothesis_tests(SR_brms, SR_data) 
saveRDS(tests2 %>% clean_table(), file = "output/tab_S7.rds")
tests2 %>% pander(split.cell = 40, split.table = Inf)
SD Mean_1 Mean_2 Estimate Est.Error CI.Lower CI.Upper Posterior_probability Evidence_ratio Notable
SD-5 Neither, 0 Father, 0 0.3101 8.112 -15.69 15.94 0.482 1.075
SD-72 Neither, 0 Father, 0 -7.195 6.051 -19.08 4.94 0.1136 7.801
SD-Mad Neither, 0 Father, 0 -11.08 6.007 -23.02 0.7857 0.03387 28.52 *
SD-5 Neither, 0 Mother, 0 -2.489 5.691 -13.67 8.988 0.323 2.096
SD-72 Neither, 0 Mother, 0 -3.226 5.698 -14.51 7.874 0.2835 2.527
SD-Mad Neither, 0 Mother, 0 -0.8937 5.896 -12.41 11.04 0.437 1.288
SD-5 Mother, 0 Father, 0 2.799 8.09 -13.38 18.37 0.3619 1.763
SD-72 Mother, 0 Father, 0 -3.968 6.293 -16.12 8.469 0.2602 2.842
SD-Mad Mother, 0 Father, 0 -10.19 6.341 -22.57 2.672 0.05613 16.82
SD-5 Mother, 1 Father, 1 1.525 5.71 -9.865 12.97 0.3913 1.556
SD-72 Mother, 1 Father, 1 -5.556 5.746 -16.99 5.818 0.1593 5.279
SD-Mad Mother, 1 Father, 1 -18.32 5.736 -29.69 -7.083 0.001375 726.3 *
SD-5 Mother, 0 Mother, 1 -5.072 5.797 -16.62 6.16 0.1869 4.351
SD-72 Mother, 0 Mother, 1 3.892 5.955 -7.87 15.7 0.2506 2.99
SD-Mad Mother, 0 Mother, 1 6.743 6.041 -4.966 18.59 0.128 6.812
SD-5 Father, 0 Father, 1 -6.346 8.066 -22.36 9.867 0.2083 3.802
SD-72 Father, 0 Father, 1 2.304 5.963 -9.459 13.82 0.343 1.915
SD-Mad Father, 0 Father, 1 -1.387 5.949 -13.18 10.16 0.4101 1.438
SD-Mad Both parents, 1 Both parents, 2 5.197 5.545 -5.9 16.09 0.1635 5.116

Analysis of adult female fitness

Run a Bayesian negative binomial generalised linear mixed model

if(!file.exists("data/model_output/female_brms.rds")){
  female_brms <-  brm(larvae_produced ~ number_of_laying_females + SD * copies_of_SD * parent_with_SD, 
                    family = negbinomial,
                    data = females, iter = 4000, chains = 4, cores = nCores,
                    control = list(adapt_delta = 0.99, max_treedepth = 15),
                    prior = prior(normal(0, 5), class = "b"))
  
  saveRDS(female_brms, "data/model_output/female_brms.rds")
} else female_brms <- readRDS("data/model_output/female_brms.rds")

Check a diagnostic plot of the model

This plot shows the frequency distribution of the original data (black line), along with predicted data that were computed using 10 random samples from the posterior of the model parameters. The posterior predicted values closely match the originals, suggesting that the model is approximating the data reasonably well.

pp_check(female_brms)

Version Author Date
437b6d7 lukeholman 2019-06-11
8d92589 lukeholman 2019-03-12
6b03f9c lukeholman 2019-03-12

Inspect the model summary

This summary shows the median, error, and 95% CIs for the posterior for all the fixed effects. This table is difficult to interpret and not very informative, but is presented here for completeness.

summarise_brms(female_brms)
Covariate Estimate Est.Error l-95% CI u-95% CI Notable
Intercept 3.97 0.34 3.32 4.64 *
number_of_laying_females 0.18 0.06 0.06 0.30 *
SDSD-5 0.04 3.00 -5.88 5.84
SDSD-72 0.07 2.96 -5.62 5.75
SDSD-Mad 0.09 3.04 -5.93 6.02
copies_of_SD1 0.06 3.39 -6.59 6.60
copies_of_SD2 -0.08 4.31 -8.50 8.48
parent_with_SDFather 0.01 3.02 -5.86 5.94
parent_with_SDMother 0.20 3.03 -5.77 6.12
parent_with_SDBothparents -0.19 3.02 -6.19 5.73
SDSD-5:copies_of_SD1 -0.03 3.27 -6.46 6.38
SDSD-72:copies_of_SD1 0.09 3.26 -6.28 6.39
SDSD-Mad:copies_of_SD1 0.17 3.31 -6.19 6.65
SDSD-5:copies_of_SD2 0.03 4.88 -9.70 9.71
SDSD-72:copies_of_SD2 0.00 5.00 -9.67 9.74
SDSD-Mad:copies_of_SD2 -0.12 4.43 -8.81 8.62
SDSD-5:parent_with_SDFather -0.13 3.28 -6.55 6.27
SDSD-72:parent_with_SDFather 0.12 3.28 -6.38 6.59
SDSD-Mad:parent_with_SDFather -0.03 3.28 -6.39 6.44
SDSD-5:parent_with_SDMother 0.19 3.26 -6.31 6.46
SDSD-72:parent_with_SDMother -0.01 3.25 -6.41 6.43
SDSD-Mad:parent_with_SDMother 0.24 3.35 -6.43 6.69
SDSD-5:parent_with_SDBothparents 0.08 3.20 -6.25 6.23
SDSD-72:parent_with_SDBothparents -0.02 3.32 -6.56 6.49
SDSD-Mad:parent_with_SDBothparents -0.35 3.34 -6.82 6.28
copies_of_SD1:parent_with_SDFather -0.01 3.36 -6.55 6.45
copies_of_SD2:parent_with_SDFather -0.06 5.06 -9.93 9.53
copies_of_SD1:parent_with_SDMother -0.18 3.34 -6.74 6.51
copies_of_SD2:parent_with_SDMother -0.01 4.99 -9.67 9.78
copies_of_SD1:parent_with_SDBothparents 0.14 3.29 -6.31 6.50
copies_of_SD2:parent_with_SDBothparents -0.01 4.30 -8.55 8.40
SDSD-5:copies_of_SD1:parent_with_SDFather 0.13 3.31 -6.35 6.54
SDSD-72:copies_of_SD1:parent_with_SDFather -0.10 3.22 -6.42 6.30
SDSD-Mad:copies_of_SD1:parent_with_SDFather 0.04 3.38 -6.58 6.66
SDSD-5:copies_of_SD2:parent_with_SDFather -0.02 5.06 -10.04 9.88
SDSD-72:copies_of_SD2:parent_with_SDFather 0.03 5.05 -9.86 9.96
SDSD-Mad:copies_of_SD2:parent_with_SDFather 0.02 4.94 -9.61 9.85
SDSD-5:copies_of_SD1:parent_with_SDMother -0.22 3.31 -6.65 6.15
SDSD-72:copies_of_SD1:parent_with_SDMother 0.22 3.33 -6.21 6.79
SDSD-Mad:copies_of_SD1:parent_with_SDMother -0.18 3.38 -6.81 6.37
SDSD-5:copies_of_SD2:parent_with_SDMother 0.02 4.98 -9.75 10.09
SDSD-72:copies_of_SD2:parent_with_SDMother 0.00 4.94 -9.60 9.69
SDSD-Mad:copies_of_SD2:parent_with_SDMother 0.01 5.02 -9.79 9.80
SDSD-5:copies_of_SD1:parent_with_SDBothparents -0.12 3.25 -6.41 6.20
SDSD-72:copies_of_SD1:parent_with_SDBothparents 0.02 3.32 -6.32 6.66
SDSD-Mad:copies_of_SD1:parent_with_SDBothparents 0.34 3.33 -6.18 6.87
SDSD-5:copies_of_SD2:parent_with_SDBothparents 0.02 5.00 -9.75 9.82
SDSD-72:copies_of_SD2:parent_with_SDBothparents -0.05 4.96 -9.70 9.57
SDSD-Mad:copies_of_SD2:parent_with_SDBothparents -0.20 4.42 -8.76 8.64

Hypothesis testing

Table S8: The results of hypothesis tests computed using the model of female fitness in Experiment 1. Each row gives the posterior estimate of a difference in means, such that the estimate is positive if mean 1 is larger than mean 2, and negative otherwise (expressed as the number of offspring produced). The mean 1 and mean 2 columns list the parent which had SD (mother, father, or both), followed by the number of SD alleles present in the offspring (0, 1 or 2). The Posterior probability column gives the probability that the mean with the smaller point estimate is actually larger than the other mean, analagously to a one-tailed p-value. Asterisks highlight rows where the posterior probability is less than 0.05.

tests3 <- hypothesis_tests(female_brms, females, divisor = 5) 
tab_S8 <- tests3 %>% clean_table()
saveRDS(tab_S8, file = "output/tab_S8.rds")
tab_S8 %>% pander(split.cell = 40, split.table = Inf)
SD Comparison Difference Error Posterior probability
SD-5 Neither, 0 - Father, 0 -1.9 (-14.7 to 12.3) 6.9 0.3684
SD-72 Neither, 0 - Father, 0 5.8 ( -6.5 to 18.5) 6.2 0.1609
SD-Mad Neither, 0 - Father, 0 1.7 ( -9.9 to 12.7) 5.8 0.3711
SD-5 Neither, 0 - Mother, 0 14.4 ( 0.4 to 28.6) 7.0 0.02187
SD-72 Neither, 0 - Mother, 0 7.9 ( -4.5 to 20.5) 6.4 0.1049
SD-Mad Neither, 0 - Mother, 0 18.7 ( 3.8 to 34.7) 7.8 0.005875
SD-5 Mother, 0 - Father, 0 -16.4 (-31.0 to -1.0) 7.6 0.01862
SD-72 Mother, 0 - Father, 0 -2.0 (-14.7 to 10.9) 6.4 0.3709
SD-Mad Mother, 0 - Father, 0 -17.0 (-32.6 to -3.2) 7.5 0.008625
SD-5 Mother, 1 - Father, 1 0.1 (-10.8 to 11.5) 5.7 0.4999
SD-72 Mother, 1 - Father, 1 -8.2 (-22.8 to 5.5) 7.2 0.1254
SD-Mad Mother, 1 - Father, 1 -3.4 (-17.7 to 10.5) 7.2 0.3214
SD-5 Mother, 0 - Mother, 1 -12.5 (-26.1 to 0.2) 6.6 0.02687
SD-72 Mother, 0 - Mother, 1 7.5 ( -6.9 to 22.5) 7.4 0.1456
SD-Mad Mother, 0 - Mother, 1 -5.4 (-22.3 to 10.9) 8.5 0.2546
SD-5 Father, 0 - Father, 1 3.9 (-10.1 to 16.9) 6.8 0.2641
SD-72 Father, 0 - Father, 1 1.4 (-11.4 to 13.9) 6.3 0.412
SD-Mad Father, 0 - Father, 1 8.2 ( -3.4 to 20.4) 6.0 0.07975
SD-Mad Both parents, 1 - Both parents, 2 -23.4 (-33.1 to -15.0) 4.6 0

Analysis of adult male fitness

Run a Bayesian binomial generalised linear mixed model

if(!file.exists("data/model_output/male_brms.rds")){
  male_brms <-  brm(focal.male.fitness | trials(total) ~ SD * copies_of_SD * parent_with_SD + (1 | vial), 
                    family = binomial,
                    data = males, iter = 4000, chains = 4, cores = nCores,
                    control = list(adapt_delta = 0.99, max_treedepth = 15),
                    prior = prior(normal(0, 5), class = "b"))
  
  saveRDS(male_brms, "data/model_output/male_brms.rds")
} else male_brms <- readRDS("data/model_output/male_brms.rds")

Check a diagnostic plot of the model

This plot shows the frequency distribution of the original data (black line), along with predicted data that were computed using 10 random samples from the posterior of the model parameters. The posterior predicted values closely match the originals, suggesting that the model is approximating the data reasonably well.

pp_check(male_brms)

Version Author Date
437b6d7 lukeholman 2019-06-11
8d92589 lukeholman 2019-03-12
6b03f9c lukeholman 2019-03-12

Inspect the model summary

This summary shows the median, error, and 95% CIs for the posterior for all the fixed effects. This table is difficult to interpret and not very informative, but is presented here for completeness.

summarise_brms(male_brms)
Covariate Estimate Est.Error l-95% CI u-95% CI Notable
Intercept 1.60 0.44 0.75 2.48 *
SDSD-5 -0.30 3.08 -6.24 5.71
SDSD-72 0.21 3.02 -5.72 6.12
SDSD-Mad 0.09 3.03 -5.81 5.89
copies_of_SD1 -0.53 3.31 -6.99 5.93
copies_of_SD2 -1.11 4.30 -9.86 7.50
parent_with_SDFather 0.00 3.03 -5.93 5.90
parent_with_SDMother 0.12 3.04 -5.83 6.05
parent_with_SDBothparents -0.22 3.06 -6.32 5.83
SDSD-5:copies_of_SD1 -1.43 3.36 -8.01 5.11
SDSD-72:copies_of_SD1 0.25 3.34 -6.44 6.75
SDSD-Mad:copies_of_SD1 0.78 3.37 -5.81 7.39
SDSD-5:copies_of_SD2 -0.04 5.11 -10.10 10.05
SDSD-72:copies_of_SD2 -0.01 5.05 -9.92 9.53
SDSD-Mad:copies_of_SD2 -1.07 4.37 -9.68 7.51
SDSD-5:parent_with_SDFather -0.29 3.37 -6.78 6.42
SDSD-72:parent_with_SDFather 0.52 3.29 -5.85 6.99
SDSD-Mad:parent_with_SDFather -0.22 3.30 -6.73 6.23
SDSD-5:parent_with_SDMother 0.69 3.32 -5.96 7.19
SDSD-72:parent_with_SDMother -0.26 3.34 -6.80 6.32
SDSD-Mad:parent_with_SDMother -0.18 3.33 -6.59 6.44
SDSD-5:parent_with_SDBothparents -0.59 3.49 -7.36 6.40
SDSD-72:parent_with_SDBothparents -0.16 3.38 -6.75 6.45
SDSD-Mad:parent_with_SDBothparents 0.62 3.31 -5.94 7.15
copies_of_SD1:parent_with_SDFather -0.44 3.29 -6.91 6.05
copies_of_SD2:parent_with_SDFather 0.10 4.93 -9.52 9.66
copies_of_SD1:parent_with_SDMother -0.16 3.28 -6.48 6.36
copies_of_SD2:parent_with_SDMother 0.03 5.06 -9.99 9.89
copies_of_SD1:parent_with_SDBothparents 0.01 3.38 -6.63 6.59
copies_of_SD2:parent_with_SDBothparents -0.94 4.25 -9.33 7.32
SDSD-5:copies_of_SD1:parent_with_SDFather -1.27 3.32 -7.70 5.33
SDSD-72:copies_of_SD1:parent_with_SDFather -0.14 3.32 -6.74 6.29
SDSD-Mad:copies_of_SD1:parent_with_SDFather 1.24 3.36 -5.35 7.87
SDSD-5:copies_of_SD2:parent_with_SDFather 0.02 5.09 -10.01 10.11
SDSD-72:copies_of_SD2:parent_with_SDFather 0.03 4.89 -9.63 9.51
SDSD-Mad:copies_of_SD2:parent_with_SDFather 0.01 5.05 -9.95 9.91
SDSD-5:copies_of_SD1:parent_with_SDMother -1.02 3.37 -7.61 5.37
SDSD-72:copies_of_SD1:parent_with_SDMother 0.60 3.36 -6.16 7.09
SDSD-Mad:copies_of_SD1:parent_with_SDMother 0.37 3.34 -6.29 6.77
SDSD-5:copies_of_SD2:parent_with_SDMother 0.00 5.01 -9.93 9.70
SDSD-72:copies_of_SD2:parent_with_SDMother -0.03 5.08 -9.85 10.14
SDSD-Mad:copies_of_SD2:parent_with_SDMother -0.01 5.03 -9.98 10.05
SDSD-5:copies_of_SD1:parent_with_SDBothparents 0.94 3.43 -6.00 7.52
SDSD-72:copies_of_SD1:parent_with_SDBothparents -0.15 3.49 -6.95 6.90
SDSD-Mad:copies_of_SD1:parent_with_SDBothparents -0.65 3.34 -7.14 5.87
SDSD-5:copies_of_SD2:parent_with_SDBothparents -0.01 5.05 -9.75 9.83
SDSD-72:copies_of_SD2:parent_with_SDBothparents -0.04 4.98 -9.92 9.61
SDSD-Mad:copies_of_SD2:parent_with_SDBothparents -0.93 4.29 -9.20 7.43

Hypothesis testing

Table S9: The results of hypothesis tests computed using the model of male fitness in Experiment 1. Each row gives the posterior estimate of a difference in means, such that the estimate is positive if mean 1 is larger than mean 2, and negative otherwise (expressed in % offspring sired). The mean 1 and mean 2 columns list the parent which had SD (mother, father, or both), followed by the number of SD alleles present in the offspring (0, 1 or 2). The Posterior probability column gives the probability that the mean with the smaller point estimate is actually larger than the other mean, analagously to a one-tailed p-value. Asterisks highlight rows where the posterior probability is less than 0.05.

tests4 <- hypothesis_tests(male_brms, males) 
tab_S9 <- tests4 %>% clean_table()
saveRDS(tab_S9, file = "output/tab_S9.rds")
tab_S9 %>% pander(split.cell = 40, split.table = Inf)
SD Comparison Difference Error Posterior probability
SD-5 Neither, 0 - Father, 0 -10.9 (-42.5 to 13.9) 14.6 0.2375
SD-72 Neither, 0 - Father, 0 8.3 ( -4.4 to 23.7) 7.2 0.1099
SD-Mad Neither, 0 - Father, 0 -1.7 (-19.4 to 15.5) 8.9 0.4201
SD-5 Neither, 0 - Mother, 0 6.2 ( -7.6 to 22.2) 7.5 0.1953
SD-72 Neither, 0 - Mother, 0 0.9 (-16.6 to 18.2) 8.6 0.4524
SD-Mad Neither, 0 - Mother, 0 0.2 (-18.0 to 18.2) 9.2 0.483
SD-5 Mother, 0 - Father, 0 -17.2 (-47.4 to 5.8) 13.9 0.09125
SD-72 Mother, 0 - Father, 0 7.3 ( -4.7 to 22.6) 7.0 0.1331
SD-Mad Mother, 0 - Father, 0 -2.0 (-19.3 to 16.1) 8.9 0.3945
SD-5 Mother, 1 - Father, 1 -20.0 (-39.2 to -5.5) 8.8 0.003375
SD-72 Mother, 1 - Father, 1 -4.8 (-19.9 to 9.6) 7.4 0.2538
SD-Mad Mother, 1 - Father, 1 3.8 ( -5.1 to 13.8) 4.7 0.2009
SD-5 Mother, 0 - Mother, 1 -61.6 (-77.7 to -41.0) 9.4 0
SD-72 Mother, 0 - Mother, 1 2.2 (-12.7 to 18.4) 7.8 0.3987
SD-Mad Mother, 0 - Mother, 1 5.7 ( -8.5 to 23.0) 7.8 0.229
SD-5 Father, 0 - Father, 1 -64.5 (-85.6 to -34.6) 13.5 0
SD-72 Father, 0 - Father, 1 -9.9 (-23.9 to 2.2) 6.7 0.055
SD-Mad Father, 0 - Father, 1 11.5 ( -0.6 to 26.7) 6.9 0.03213
SD-Mad Both parents, 1 - Both parents, 2 -70.6 (-82.6 to -54.2) 7.2 0

Making Figure 1

make_plot <- function(model, data, ylab, ylim = NULL){
  new <- data %>% 
    distinct(SD, copies_of_SD, MotherHasSD, FatherHasSD, SD_from_mum, SD_from_dad, mum_genotype, dad_genotype, parent_with_SD) %>% 
    mutate(total = 100, number_of_laying_females = 5, larval_density = 0)
  
  dots <- data.frame(new, fitted(model, newdata = new, re_formula = NA)) %>%
    mutate(parent_with_SD = ifelse(MotherHasSD == 1, "Mother", "Father"),
           parent_with_SD = replace(parent_with_SD, MotherHasSD == 1 & FatherHasSD == 1, "Both")) %>%
    mutate(parent_with_SD = factor(parent_with_SD, c("Father", "Mother", "Both")))
  
  inner_bar <- data.frame(new, fitted(model, newdata = new, re_formula = NA, probs = c(0.25, 0.75))) %>%
    mutate(parent_with_SD = ifelse(MotherHasSD == 1, "Mother", "Father"),
           parent_with_SD = replace(parent_with_SD, MotherHasSD == 1 & FatherHasSD == 1, "Both")) %>%
    mutate(parent_with_SD = factor(parent_with_SD, c("Father", "Mother", "Both")))
  
  lowerCI <- dots$Q2.5[dots$MotherHasSD == 0 & dots$FatherHasSD == 0]
  upperCI <- dots$Q97.5[dots$MotherHasSD == 0 & dots$FatherHasSD == 0]
  dots <- dots %>% filter(!(MotherHasSD == 0 & FatherHasSD == 0)) %>%
    left_join(inner_bar %>% select(-Estimate, -Est.Error))
  
  if("number_of_laying_females" %in% names(data)){
    dots$Estimate <- dots$Estimate / 5
    dots$Q2.5 <- dots$Q2.5 / 5
    dots$Q97.5 <- dots$Q97.5 / 5
    dots$Q25 <- dots$Q25 / 5
    dots$Q75 <- dots$Q75 / 5
    lowerCI <- lowerCI / 5
    upperCI <- upperCI / 5
  }
  
  pal <- got(n=4, alpha = 1, begin = 0.2, end = 0.8, direction = -1, option = "tully")
  cols <- c("0" = pal[1], "0 or 1" = pal[2], "1" = pal[3], "2" = pal[4])

  
  pd <- position_dodge(0.45)
  plot <- dots %>%
    mutate(copies_of_SD = factor(copies_of_SD, levels = c("0", "0 or 1", "1", "2"))) %>%
    ggplot(aes(parent_with_SD, Estimate, colour = copies_of_SD, fill = copies_of_SD)) + 
    geom_errorbar(aes(ymin = Q2.5, ymax = Q97.5), width = 0, position = pd, alpha = 0) + # invisible layer - placeholder for scales
    annotate("rect", xmin = -Inf, xmax = Inf, ymin = lowerCI, ymax = upperCI, alpha = 0.2, fill = "skyblue") + 
    geom_errorbar(aes(ymin = Q2.5, ymax = Q97.5), width = 0, position = pd, size = 1.2) + 
    geom_errorbar(aes(ymin = Q25, ymax = Q75), width = 0, position = pd, size = 3) + 
    geom_point(position = pd, size = 2, colour = "black", pch = 21) + 
    # scale_fill_got(option = "Lady Margaery", name = "Copies of SD", discrete = TRUE) + 
    # scale_colour_got(option = "Lady Margaery", name = "Copies of SD", discrete = TRUE) + 
    
    scale_colour_manual(name = "Copies of SD", values = cols,
                        drop = FALSE, 
                        aesthetics = c("colour", "fill")) +

    facet_wrap(~SD) + 
    xlab("Parent carrying SD") + ylab(ylab) + theme_hc() + theme(axis.ticks.y = element_blank())
  
  if(!is.null(ylim))  plot <- plot + scale_y_continuous(limits = ylim) 
  plot

}

make_expt1_multiplot <- function(...) {
    plots <- list(...)
    g <- ggplotGrob(plots[[1]] + theme(legend.position="top"))$grobs
    legend <- g[[which(sapply(g, function(x) x$name) == "guide-box")]]
    lheight <- sum(legend$height)
    plots <- lapply(plots, function(x) ggplotGrob(x + theme(legend.position = "none")))
    
    p <- gtable_cbind(gtable_rbind(plots[[1]], plots[[3]]), 
                      gtable_rbind(plots[[2]], plots[[4]]))
    
    grid.arrange(legend, p,
                 nrow = 2, 
                 heights = unit.c(lheight, unit(1, "npc") - lheight),
                 bottom = "Parent carrying SD")
}

fig1 <- make_expt1_multiplot(
  make_plot(larvae_brms, larvae, "Juvenile fitness (% survival)") + xlab(NULL) + theme(axis.text.x = element_blank(), axis.ticks.x = element_blank()),
  make_plot(SR_brms, SR_data, "Adult sex ratio (% sons)") + xlab(NULL) + theme(axis.text.x = element_blank(), axis.ticks.x = element_blank()),
  make_plot(female_brms, females, "Adult female fitness (number of progeny)", c(0, 62)) + xlab(NULL),
  make_plot(male_brms, males, "Adult male fitness (% offspring sired)", c(0, 100)) + xlab(NULL)
)

Version Author Date
3244bce lukeholman 2019-06-11
437b6d7 lukeholman 2019-06-11
6b03f9c lukeholman 2019-03-12
ggsave(fig1, file = "figures/fig1.pdf", height = 7, width = 11)

Figure 1: Posterior estimates of the group means for the four different response variables in Experiment 1, for each type of cross (x-axis), SD variant (panels), and genotype (colours). Juvenile fitness was measured as % L1 larva-to-adult survival, adult sex ratio refers to the number of males and females among the individuals that reached adulthood, female fitness is the estimated number of progeny produced per female, and male fitness as the siring success relative to competitior males. The thicker inner bar shows the region containing 50% of the posterior, the outer bar covers 95% of the posterior, and the circle marks the median. Tables S6-S9 give the accompanying statistcal results. Points labelled as carrying ``0 or 1’’ SD allele refer to cases where the genotype of the offspring could not be ascertained due to the brown eye marker not being expressed in larvae; most of these individuals probably carried 1 SD allele because of segregation distortion.

Making Table 1

Table 1: List of the all the notable differences between groups in Experiment 1 (posterior probability <0.05; see Tables S6-S9 for the remaining results). For each contast, we give the parent carrying SD (neither, mother, father, or both) and the number of SD alleles carried by the offspring. The difference in means is expressed in the original units, i.e. % larvae surviving, % male larvae, number of larvae produced, or % offspring sired, and the parentheses give 95% credible intervals. The difference is positive when the first-listed mean is larger than the second-listed mean, and negative otherwise. The posterior probability (p) has a similar interpretation to a one-tailed p-value.

table1 <- bind_rows(tests1 %>% mutate(Trait = "Larval survival"),
                    tests2 %>% mutate(Trait = "Sex ratio"),
                    tests3 %>% mutate(Trait = "Female fitness"),
                    tests4 %>% mutate(Trait = "Male fitness")) %>% 
  arrange(SD, Trait, Posterior_probability) %>%
  filter(Notable == "*") %>% clean_table() %>% 
  select(SD, everything())

write_csv(table1 %>% rename(p = `Posterior probability`), path = "figures/Table1.csv")
table1 %>% pander(split.cell = 50, split.table = Inf) 
SD Trait Comparison Difference Error Posterior probability
SD-5 Female fitness Mother, 0 - Father, 0 -16.4 (-31.0 to -1.0) 7.6 0.01862
SD-5 Female fitness Neither, 0 - Mother, 0 14.4 ( 0.4 to 28.6) 7.0 0.02187
SD-5 Female fitness Mother, 0 - Mother, 1 -12.5 (-26.1 to 0.2) 6.6 0.02687
SD-5 Larval survival Both parents, 0 or 1 - Both parents, 2 -77.1 (-87.8 to -62.2) 6.5 0
SD-5 Male fitness Mother, 0 - Mother, 1 -61.6 (-77.7 to -41.0) 9.4 0
SD-5 Male fitness Father, 0 - Father, 1 -64.5 (-85.6 to -34.6) 13.5 0
SD-5 Male fitness Mother, 1 - Father, 1 -20.0 (-39.2 to -5.5) 8.8 0.003375
SD-72 Larval survival Both parents, 0 or 1 - Both parents, 2 -77.5 (-87.8 to -63.6) 6.1 0
SD-Mad Female fitness Both parents, 1 - Both parents, 2 -23.4 (-33.1 to -15.0) 4.6 0
SD-Mad Female fitness Neither, 0 - Mother, 0 18.7 ( 3.8 to 34.7) 7.8 0.005875
SD-Mad Female fitness Mother, 0 - Father, 0 -17.0 (-32.6 to -3.2) 7.5 0.008625
SD-Mad Male fitness Both parents, 1 - Both parents, 2 -70.6 (-82.6 to -54.2) 7.2 0
SD-Mad Male fitness Father, 0 - Father, 1 11.5 ( -0.6 to 26.7) 6.9 0.03213
SD-Mad Sex ratio Mother, 1 - Father, 1 -18.3 (-29.7 to -7.1) 5.7 0.001375
SD-Mad Sex ratio Neither, 0 - Father, 0 -11.1 (-23.0 to 0.8) 6.0 0.03387

Experiment 2

Run a Bayesian binomial generalised linear mixed model

# Function to simular the missing genotypes of embryos that died, assuming fair meiosis in SD mothers and k = k in SD fathers
simulate_expt2_data <- function(expt2_dataset, k){
  
  have_SD_mum <- which(expt2_dataset$SD_parent == "mum")
  have_SD_dad <- which(expt2_dataset$SD_parent == "dad")
  
  expt2_dataset$SD_embryos[have_SD_mum] <- with(expt2_dataset[have_SD_mum, ], 
                                                SD_offspring + rbinom(length(SD_embryos), Dead_offspring, 0.5))
  
  expt2_dataset$SD_embryos[have_SD_dad] <- with(expt2_dataset[have_SD_dad, ], 
                                                SD_offspring + rbinom(length(SD_embryos), Dead_offspring, k))
  
  expt2_dataset <- expt2_dataset %>%
    mutate(SD_embryos = as.numeric(SD_embryos),
           CyO_embryos = starting_number_focal_sex - SD_embryos) %>% 
    select(-starting_number_focal_sex, -Dead_offspring) 
  
  expt2_dataset %>% 
    select(sex_of_embryos, SD, SD_parent, vial, block, SD_offspring, CyO_offspring) %>%
    tidyr::gather(genotype, num_survivors, SD_offspring, CyO_offspring) %>%
    mutate(genotype = gsub("_offspring", "", genotype)) %>%
    left_join(expt2_dataset %>% 
                select(sex_of_embryos, SD, SD_parent, vial, block, SD_embryos, CyO_embryos) %>%
                tidyr::gather(genotype, total_number, SD_embryos, CyO_embryos) %>%
                mutate(genotype = gsub("_embryos", "", genotype))) %>%
    arrange(vial)
}

# Simulate some data and run the Bayesian GLMM
model_simulated_data <- function(expt2_dataset, k){
  
  # First re-format teh data and simulate the missing embryo genotypes using stated value of k
  dat <- simulate_expt2_data(expt2_dataset, k)

  # Now run the model
  brm(num_survivors  | trials(total_number) ~ genotype * sex_of_embryos * SD * SD_parent + (1 | vial), 
      data = dat, family = "binomial",
      chains = 4, cores = nCores, seed = 1,
      prior = prior(normal(0, 5), class = "b"),
      control = list(adapt_delta = 0.9999, max_treedepth = 15))  
  
}

# Do the same model for k = 0.5, 0.6 and 0.7
if(!file.exists("data/model_output/expt2_brms.rds")){
  model_list <- list(
    model_simulated_data(expt2, 0.5),
    model_simulated_data(expt2, 0.6),
    model_simulated_data(expt2, 0.7))
  
  saveRDS(model_list, "data/model_output/expt2_brms.rds")
} else model_list <- readRDS("data/model_output/expt2_brms.rds")

Check a diagnostic plot of the model

This plot shows the frequency distribution of the original data (black line), along with predicted data that were computed using 10 random samples from the posterior of the model parameters. The posterior predicted values closely match the originals, suggesting that the model is approximating the data reasonably well. There are two plots for this model because we used a multivariate model that jointly modelled two response variables: the survival of SD larvae and of CyO larvae.

model_list[[1]]
 Family: binomial 
  Links: mu = logit 
Formula: num_survivors | trials(total_number) ~ genotype * sex_of_embryos * SD * SD_parent + (1 | vial) 
   Data: dat (Number of observations: 422) 
Samples: 4 chains, each with iter = 2000; warmup = 1000; thin = 1;
         total post-warmup samples = 4000

Group-Level Effects: 
~vial (Number of levels: 211) 
              Estimate Est.Error l-95% CI u-95% CI Eff.Sample Rhat
sd(Intercept)     0.63      0.05     0.54     0.74       1621 1.00

Population-Level Effects: 
                                                    Estimate Est.Error
Intercept                                               1.80      0.21
genotypeSD                                             -0.10      0.20
sex_of_embryosmale                                     -0.19      0.29
SDSDM72                                                 0.16      0.30
SDSDMMad                                                0.10      0.35
SD_parentmum                                           -0.19      0.28
genotypeSD:sex_of_embryosmale                          -0.17      0.27
genotypeSD:SDSDM72                                      0.07      0.29
genotypeSD:SDSDMMad                                     0.16      0.33
sex_of_embryosmale:SDSDM72                             -0.34      0.42
sex_of_embryosmale:SDSDMMad                             0.28      0.50
genotypeSD:SD_parentmum                                -0.21      0.26
sex_of_embryosmale:SD_parentmum                        -0.45      0.39
SDSDM72:SD_parentmum                                    0.30      0.41
SDSDMMad:SD_parentmum                                   0.44      0.45
genotypeSD:sex_of_embryosmale:SDSDM72                   0.34      0.40
genotypeSD:sex_of_embryosmale:SDSDMMad                 -0.09      0.47
genotypeSD:sex_of_embryosmale:SD_parentmum              0.32      0.35
genotypeSD:SDSDM72:SD_parentmum                         0.49      0.38
genotypeSD:SDSDMMad:SD_parentmum                       -0.34      0.42
sex_of_embryosmale:SDSDM72:SD_parentmum                 0.16      0.54
sex_of_embryosmale:SDSDMMad:SD_parentmum               -0.40      0.61
genotypeSD:sex_of_embryosmale:SDSDM72:SD_parentmum     -0.54      0.51
genotypeSD:sex_of_embryosmale:SDSDMMad:SD_parentmum     0.62      0.57
                                                    l-95% CI u-95% CI
Intercept                                               1.40     2.21
genotypeSD                                             -0.49     0.28
sex_of_embryosmale                                     -0.76     0.37
SDSDM72                                                -0.42     0.74
SDSDMMad                                               -0.58     0.77
SD_parentmum                                           -0.74     0.37
genotypeSD:sex_of_embryosmale                          -0.70     0.36
genotypeSD:SDSDM72                                     -0.49     0.65
genotypeSD:SDSDMMad                                    -0.49     0.82
sex_of_embryosmale:SDSDM72                             -1.14     0.52
sex_of_embryosmale:SDSDMMad                            -0.70     1.25
genotypeSD:SD_parentmum                                -0.71     0.32
sex_of_embryosmale:SD_parentmum                        -1.20     0.33
SDSDM72:SD_parentmum                                   -0.50     1.10
SDSDMMad:SD_parentmum                                  -0.41     1.31
genotypeSD:sex_of_embryosmale:SDSDM72                  -0.45     1.13
genotypeSD:sex_of_embryosmale:SDSDMMad                 -1.00     0.83
genotypeSD:sex_of_embryosmale:SD_parentmum             -0.36     1.01
genotypeSD:SDSDM72:SD_parentmum                        -0.26     1.23
genotypeSD:SDSDMMad:SD_parentmum                       -1.18     0.48
sex_of_embryosmale:SDSDM72:SD_parentmum                -0.93     1.24
sex_of_embryosmale:SDSDMMad:SD_parentmum               -1.58     0.78
genotypeSD:sex_of_embryosmale:SDSDM72:SD_parentmum     -1.52     0.46
genotypeSD:sex_of_embryosmale:SDSDMMad:SD_parentmum    -0.47     1.73
                                                    Eff.Sample Rhat
Intercept                                                  957 1.00
genotypeSD                                                1282 1.01
sex_of_embryosmale                                         988 1.00
SDSDM72                                                    894 1.00
SDSDMMad                                                  1056 1.00
SD_parentmum                                               800 1.00
genotypeSD:sex_of_embryosmale                             1272 1.01
genotypeSD:SDSDM72                                        1383 1.00
genotypeSD:SDSDMMad                                       1524 1.00
sex_of_embryosmale:SDSDM72                                 933 1.00
sex_of_embryosmale:SDSDMMad                               1144 1.00
genotypeSD:SD_parentmum                                   1323 1.00
sex_of_embryosmale:SD_parentmum                            863 1.00
SDSDM72:SD_parentmum                                       820 1.00
SDSDMMad:SD_parentmum                                      937 1.01
genotypeSD:sex_of_embryosmale:SDSDM72                     1453 1.00
genotypeSD:sex_of_embryosmale:SDSDMMad                    1571 1.00
genotypeSD:sex_of_embryosmale:SD_parentmum                1259 1.00
genotypeSD:SDSDM72:SD_parentmum                           1308 1.00
genotypeSD:SDSDMMad:SD_parentmum                          1440 1.00
sex_of_embryosmale:SDSDM72:SD_parentmum                    911 1.00
sex_of_embryosmale:SDSDMMad:SD_parentmum                  1030 1.00
genotypeSD:sex_of_embryosmale:SDSDM72:SD_parentmum        1363 1.00
genotypeSD:sex_of_embryosmale:SDSDMMad:SD_parentmum       1459 1.00

Samples were drawn using sampling(NUTS). For each parameter, Eff.Sample 
is a crude measure of effective sample size, and Rhat is the potential 
scale reduction factor on split chains (at convergence, Rhat = 1).

Inspect the model summary

This summary shows the median, error, and 95% CIs for the posterior for all the fixed effects. This table is difficult to interpret and not very informative, but is presented here for completeness.

summarise_brms(model_list[[1]])
Covariate Estimate Est.Error l-95% CI u-95% CI Notable
Intercept 1.80 0.21 1.40 2.21 *
genotypeSD -0.10 0.20 -0.49 0.28
sex_of_embryosmale -0.19 0.29 -0.76 0.37
SDSD-72 0.16 0.30 -0.42 0.74
SDSD-Mad 0.10 0.35 -0.58 0.77
SD_parentmum -0.19 0.28 -0.74 0.37
genotypeSD:sex_of_embryosmale -0.17 0.27 -0.70 0.36
genotypeSD:SDSD-72 0.07 0.29 -0.49 0.65
genotypeSD:SDSD-Mad 0.16 0.33 -0.49 0.82
sex_of_embryosmale:SDSD-72 -0.34 0.42 -1.14 0.52
sex_of_embryosmale:SDSD-Mad 0.28 0.50 -0.70 1.25
genotypeSD:SD_parentmum -0.21 0.26 -0.71 0.32
sex_of_embryosmale:SD_parentmum -0.45 0.39 -1.20 0.33
SDSD-72:SD_parentmum 0.30 0.41 -0.50 1.10
SDSD-Mad:SD_parentmum 0.44 0.45 -0.41 1.31
genotypeSD:sex_of_embryosmale:SDSD-72 0.34 0.40 -0.45 1.13
genotypeSD:sex_of_embryosmale:SDSD-Mad -0.09 0.47 -1.00 0.83
genotypeSD:sex_of_embryosmale:SD_parentmum 0.32 0.35 -0.36 1.01
genotypeSD:SDSD-72:SD_parentmum 0.49 0.38 -0.26 1.23
genotypeSD:SDSD-Mad:SD_parentmum -0.34 0.42 -1.18 0.48
sex_of_embryosmale:SDSD-72:SD_parentmum 0.16 0.54 -0.93 1.24
sex_of_embryosmale:SDSD-Mad:SD_parentmum -0.40 0.61 -1.58 0.78
genotypeSD:sex_of_embryosmale:SDSD-72:SD_parentmum -0.54 0.51 -1.52 0.46
genotypeSD:sex_of_embryosmale:SDSD-Mad:SD_parentmum 0.62 0.57 -0.47 1.73

Plot the posterior predictions of the group means

new_data <- simulate_expt2_data(expt2, 0.5) %>% 
  select(sex_of_embryos, SD, SD_parent, genotype) %>% 
  distinct() %>% mutate(total_number = 100) %>%
  arrange(desc(SD_parent), SD, desc(sex_of_embryos))

plot_expt2 <- function(model){

  pd <- position_dodge(0.4)
  
  predictions <- fitted(model, 
                        summary = TRUE, 
                        newdata = new_data, re_formula = NA)
  
  inner_bar <- fitted(model, 
                      summary = TRUE, 
                      newdata = new_data, re_formula = NA,
                      probs = c(0.25, 0.75))
  
  process_preds <- function(preds){
    data.frame(new_data, preds) %>%
      rename(Offspring_type = genotype) %>%
      mutate(sex_of_embryos  = str_replace_all(sex_of_embryos, "female", "daughters"),
             sex_of_embryos  = str_replace_all(sex_of_embryos, "male", "sons"),
             Offspring_type = paste(Offspring_type, sex_of_embryos),
             SD_parent = str_replace_all(SD_parent, "dad", "SD/CyO father"),
             SD_parent = str_replace_all(SD_parent, "mum", "SD/CyO mother"))
  }

  left_join(process_preds(predictions), 
            process_preds(inner_bar)) %>% 
    ggplot(aes(Offspring_type, Estimate, colour = SD_parent, fill = SD_parent)) +
    geom_errorbar(aes(ymin = Q2.5, ymax = Q97.5), width = 0, position = pd, size = 1.3) + 
    geom_errorbar(aes(ymin = Q25, ymax = Q75), width = 0, position = pd, size = 3) + 
    geom_point(position = pd, size = 2, colour = "black",  pch = 21) +
    facet_grid(~SD) + 
    scale_colour_brewer(name = "Cross", palette = "Pastel1", direction = -1, aesthetics = c("colour", "fill")) +
    theme_hc() + 
    theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1),
          legend.position = "top", axis.ticks.y = element_blank()) +
    ylab("Juvenile fitness (% survival)") + xlab("Genotype and sex of offspring")
}

fig2 <- plot_expt2(model_list[[1]])
ggsave(fig2, file = "figures/fig2.pdf", height = 3.9, width = 7)
fig2

Version Author Date
437b6d7 lukeholman 2019-06-11
6b03f9c lukeholman 2019-03-12

Figure 2: Posterior estimates of % L1 larva-to-adult survival in Experiment 2 for each combination of offspring sex and genotype (x-axis), SD variant (panels), and cross (colours). Error bars show the 95% quantiles of the posterior. See Tables 2 and S10 for associated hypothesis tests. This plot was generated assuming fair meiosis (\(k\) = 0.5) in SD/CyO males; see Figure S1 for equivalent plots made using different assumed values of \(k\).

Check the effect of our assumption that meiosis is fair in SD/CyO males

make_expt2_multiplot <- function(...) {
    plots <- list(...)
    g <- ggplotGrob(plots[[1]] + theme(legend.position="bottom"))$grobs
    legend <- g[[which(sapply(g, function(x) x$name) == "guide-box")]]
    lheight <- sum(legend$height)
    plots <- lapply(plots, function(x) ggplotGrob(x + theme(legend.position = "none")))
    
    p <- gtable_rbind(plots[[1]], plots[[2]], plots[[3]])
    
    arrangeGrob(p, legend, 
                 nrow = 2, 
                 heights = unit.c(unit(1, "npc") - lheight, lheight))
  }

fig_S1 <- make_expt2_multiplot(
  plot_expt2(model_list[[1]]) + theme(axis.text.x = element_blank(), axis.ticks.x = element_blank()) + xlab(NULL) + ylab(NULL),
  plot_expt2(model_list[[2]]) + theme(axis.text.x = element_blank(), axis.ticks.x = element_blank()) + xlab(NULL),
  plot_expt2(model_list[[3]]) + ylab(NULL))

fig_S1 %>% ggsave(file = "output/fig_S1.pdf", height = 10, width = 9)
fig_S1 %>% grid::grid.draw()

Version Author Date
6b03f9c lukeholman 2019-03-12



Figure S1: Equivalent plots to Figure 2, under the assumption that meiosis is fair (\(k\) = 0.5, top row, same as Figure 2), slightly biased (\(k\) = 0.6, middle row), and more strongly biased (\(k\) = 0.7, bottom row). Note that the significant results for Figure 2 mostly stay the same or increase in magnitude, suggesting that they are genuine and are not sensitive to our assumptions about the data.

Hypothesis testing

hypothesis_tests_expt2 <- function(model){
  
  compare <- function(mean1, mean2, SD, Mean_1, Mean_2){
  difference <- mean2 - mean1
  prob_less <- hypothesis(data.frame(x = difference), "x < 0")$hypothesis
  prob_more <- hypothesis(data.frame(x = difference), "x > 0")$hypothesis
  output <- data.frame(
    SD, Mean_1, Mean_2,
    hypothesis(data.frame(x = difference), "x = 0")$hypothesis %>% 
    mutate(Posterior_probability = min(c(prob_less$Post.Prob, prob_more$Post.Prob)),
           Evidence_ratio = max(c(prob_less$Evid.Ratio, prob_more$Evid.Ratio)))
  )
}
  preds <- fitted(model, 
                  summary = FALSE, 
                  newdata = new_data, re_formula = NA)
  
  
  SD5_sons_mum_SD <- preds[,1]
  SD5_sons_mum_CyO <- preds[,2]
  SD5_daughters_mum_SD <- preds[,3]
  SD5_daughters_mum_CyO <- preds[,4]
  SD72_sons_mum_SD <- preds[,5]
  SD72_sons_mum_CyO <- preds[,6]
  SD72_daughters_mum_SD <- preds[,7]
  SD72_daughters_mum_CyO <- preds[,8]
  SDMad_sons_mum_SD <- preds[,9]
  SDMad_sons_mum_CyO <- preds[,10]
  SDMad_daughters_mum_SD <- preds[,11]
  SDMad_daughters_mum_CyO <- preds[,12]
  SD5_sons_dad_SD <- preds[,13]
  SD5_sons_dad_CyO <- preds[,14]
  SD5_daughters_dad_SD <- preds[,15]
  SD5_daughters_dad_CyO <- preds[,16]
  SD72_sons_dad_SD <- preds[,17]
  SD72_sons_dad_CyO <- preds[,18]
  SD72_daughters_dad_SD <- preds[,19]
  SD72_daughters_dad_CyO <- preds[,20]
  SDMad_sons_dad_SD <- preds[,21]
  SDMad_sons_dad_CyO <- preds[,22]
  SDMad_daughters_dad_SD <- preds[,23]
  SDMad_daughters_dad_CyO <- preds[,24]
  
  rbind(
    # sex effect on SD individuals when mother has SD
    compare(SD5_sons_mum_SD, SD5_daughters_mum_SD, "SD-5", "Sons, SD, mother", "Daughters, SD, mother"), 
    compare(SD72_sons_mum_SD, SD72_daughters_mum_SD, "SD-72", "Sons, SD, mother", "Daughters, SD, mother"),
    compare(SDMad_sons_mum_SD, SDMad_daughters_mum_SD, "SD-Mad", "Sons, SD, mother", "Daughters, SD, mother"),
    # sex effect on CyO individuals when mother has SD
    compare(SD5_sons_mum_CyO, SD5_daughters_mum_CyO, "SD-5", "Sons, CyO, mother", "Daughters, CyO, mother"), 
    compare(SD72_sons_mum_CyO, SD72_daughters_mum_CyO, "SD-72", "Sons, CyO, mother", "Daughters, CyO, mother"),
    compare(SDMad_sons_mum_CyO, SDMad_daughters_mum_CyO, "SD-Mad", "Sons, CyO, mother", "Daughters, CyO, mother"),
    # sex effect on SD individuals when father has SD
    compare(SD5_sons_dad_SD, SD5_daughters_dad_SD, "SD-5", "Sons, SD, father", "Daughters, SD, father"), 
    compare(SD72_sons_dad_SD, SD72_daughters_dad_SD, "SD-72", "Sons, SD, father", "Daughters, SD, father"),
    compare(SDMad_sons_dad_SD, SDMad_daughters_dad_SD, "SD-Mad", "Sons, SD, father", "Daughters, SD, father"),
    # sex effect on CyO individuals when father has SD
    compare(SD5_sons_dad_CyO, SD5_daughters_dad_CyO, "SD-5", "Sons, CyO, father", "Daughters, CyO, father"), 
    compare(SD72_sons_dad_CyO, SD72_daughters_dad_CyO, "SD-72", "Sons, CyO, father", "Daughters, CyO, father"),
    compare(SDMad_sons_dad_CyO, SDMad_daughters_dad_CyO, "SD-Mad", "Sons, CyO, father", "Daughters, CyO, father"),
    # effect of inheriting SD not CyO from mother on sons
    compare(SD5_sons_mum_CyO, SD5_sons_mum_SD, "SD-5", "Sons, CyO, mother", "Sons, SD, mother"), 
    compare(SD72_sons_mum_CyO, SD72_sons_mum_SD, "SD-72", "Sons, CyO, mother", "Sons, SD, mother"),
    compare(SDMad_sons_mum_CyO, SDMad_sons_mum_SD, "SD-Mad", "Sons, CyO, mother", "Sons, SD, mother"),
    # effect of inheriting SD not CyO from father on sons
    compare(SD5_sons_dad_CyO, SD5_sons_dad_SD, "SD-5", "Sons, CyO, father", "Sons, SD, father"),
    compare(SD72_sons_dad_CyO, SD72_sons_dad_SD, "SD-72", "Sons, CyO, father", "Sons, SD, father"),
    compare(SDMad_sons_dad_CyO, SDMad_sons_dad_SD, "SD-Mad", "Sons, CyO, father", "Sons, SD, father"),
    # effect of inheriting SD not CyO from mother on daughters
    compare(SD5_daughters_mum_CyO, SD5_daughters_mum_SD, "SD-5", "Daughters, CyO, mother", "Daughters, SD, mother"), 
    compare(SD72_daughters_mum_CyO, SD72_daughters_mum_SD, "SD-72", "Daughters, CyO, mother", "Daughters, SD, mother"),
    compare(SDMad_daughters_mum_CyO, SDMad_daughters_mum_SD, "SD-Mad", "Daughters, CyO, mother", "Daughters, SD, mother"),
    # effect of inheriting SD not CyO from father on daughters
    compare(SD5_daughters_dad_CyO, SD5_daughters_dad_SD, "SD-5", "Daughters, CyO, father", "Daughters, SD, father"), 
    compare(SD72_daughters_dad_CyO, SD72_daughters_dad_SD, "SD-72", "Daughters, CyO, father", "Daughters, SD, father"),
    compare(SDMad_daughters_dad_CyO, SDMad_daughters_dad_SD, "SD-Mad", "Daughters, CyO, father", "Daughters, SD, father"),
    # effect of cross direction for SD sons
    compare(SD5_sons_mum_SD, SD5_sons_dad_SD, "SD-5", "Sons, SD, mother", "Sons, SD, father"), 
    compare(SD72_sons_mum_SD, SD72_sons_dad_SD, "SD-72", "Sons, SD, mother", "Sons, SD, father"),
    compare(SDMad_sons_mum_SD, SDMad_sons_dad_SD, "SD-Mad", "Sons, SD, mother", "Sons, SD, father"),
    # effect of cross direction for SD daughters
    compare(SD5_daughters_mum_SD, SD5_daughters_dad_SD, "SD-5", "Daughters, SD, mother", "Daughters, SD, father"), 
    compare(SD72_daughters_mum_SD, SD72_daughters_dad_SD, "SD-72", "Daughters, SD, mother", "Daughters, SD, father"),
    compare(SDMad_daughters_mum_SD, SDMad_daughters_dad_SD, "SD-Mad", "Daughters, SD, mother", "Daughters, SD, father"),
    # effect of cross direction for CyO sons
    compare(SD5_sons_mum_CyO, SD5_sons_dad_CyO, "SD-5", "Sons, CyO, mother", "Sons, CyO, father"), 
    compare(SD72_sons_mum_CyO, SD72_sons_dad_CyO, "SD-72", "Sons, CyO, mother", "Sons, CyO, father"),
    compare(SDMad_sons_mum_CyO, SDMad_sons_dad_CyO, "SD-Mad", "Sons, CyO, mother", "Sons, CyO, father"),
    # effect of cross direction for CyO daughters
    compare(SD5_daughters_mum_CyO, SD5_daughters_dad_CyO, "SD-5", "Daughters, CyO, mother", "Daughters, CyO, father"), 
    compare(SD72_daughters_mum_CyO, SD72_daughters_dad_CyO, "SD-72", "Daughters, CyO, mother", "Daughters, CyO, father"),
    compare(SDMad_daughters_mum_CyO, SDMad_daughters_dad_CyO, "SD-Mad", "Daughters, CyO, mother", "Daughters, CyO, father")
  ) %>% 
    select(-Hypothesis, -Evid.Ratio, -Post.Prob, -Star) %>%
    mutate(Notable = ifelse(Posterior_probability < 0.05, "*", ""),
           Test = c(rep("Sex difference in survival", 12),
                    rep("Effect of offspring genotype", 12),
                    rep("Effect of parental genotype", 12))) %>% as_tibble() 
}

tests_expt2 <- hypothesis_tests_expt2(model_list[[1]]) %>% 
  arrange(SD, Posterior_probability) %>%
  mutate(Estimate = format(round(Estimate, 1), nsmall = 1),  
         Est.Error = format(round(Est.Error, 1), nsmall = 1),
         CI.Lower = format(round(CI.Lower, 1), nsmall = 1),
         CI.Upper = format(round(CI.Upper, 1), nsmall = 1), 
         Difference = paste(Estimate, " (", CI.Lower, " to ", CI.Upper, ")", sep = ""),
         Comparison = paste(Mean_1, "-", Mean_2),
         Error = Est.Error) %>%
  select(SD, Comparison, Difference, Error, Posterior_probability, Notable)

Table 2: List of the all the notable differences between groups in Experiment 2 (posterior probability $<$0.05; see Table S10 for the remaining results). For each group, we list the sex of the focal larvae, their genotype (SD or CyO), and the parent that carried SD (mother or father). The difference in means is expressed in % larvae surviving; other details are as in Table 1.

table2 <- tests_expt2 %>% 
  filter(Notable == "*") %>% select(-Notable) 

write_csv(table2 %>% rename(p = Posterior_probability), path = "figures/Table2.csv")
table2 %>% pander(split.cell = 50, split.table = Inf)
SD Comparison Difference Error Posterior_probability
SD-5 Sons, CyO, mother - Daughters, CyO, mother 10.7 ( 2.0 to 19.4) 4.5 0.00825
SD-5 Sons, CyO, mother - Sons, CyO, father 10.7 ( 1.5 to 19.8) 4.6 0.0105
SD-5 Sons, SD, mother - Sons, SD, father 9.7 ( 0.1 to 19.4) 5.0 0.02475
SD-5 Sons, SD, mother - Daughters, SD, mother 9.1 ( -0.4 to 18.7) 4.8 0.028
SD-5 Daughters, CyO, mother - Daughters, SD, mother -4.6 (-10.0 to 0.6) 2.7 0.041
SD-72 Sons, SD, mother - Daughters, SD, mother 9.9 ( 4.2 to 15.8) 3.0 5e-04
SD-72 Sons, CyO, mother - Daughters, CyO, mother 11.0 ( 4.3 to 17.9) 3.5 0.00075
SD-72 Sons, CyO, father - Daughters, CyO, father 7.0 ( -1.0 to 15.5) 4.2 0.04475
SD-Mad Sons, CyO, mother - Daughters, CyO, mother 9.4 ( 2.7 to 16.3) 3.5 0.00225
SD-Mad Daughters, CyO, mother - Daughters, SD, mother -5.4 ( -9.9 to -1.2) 2.2 0.00675
SD-Mad Sons, CyO, mother - Sons, CyO, father 7.7 ( -0.6 to 15.6) 4.1 0.0345

Table S10: Complete version of Table 2, showing all the contrasts that were tested in Experiment 2.

saveRDS(tests_expt2 %>% tidy_names() %>% select(-Notable), file = "output/tab_S10.rds")
tests_expt2 %>% pander(split.cell = 50, split.table = Inf) 
SD Comparison Difference Error Posterior_probability Notable
SD-5 Sons, CyO, mother - Daughters, CyO, mother 10.7 ( 2.0 to 19.4) 4.5 0.00825 *
SD-5 Sons, CyO, mother - Sons, CyO, father 10.7 ( 1.5 to 19.8) 4.6 0.0105 *
SD-5 Sons, SD, mother - Sons, SD, father 9.7 ( 0.1 to 19.4) 5.0 0.02475 *
SD-5 Sons, SD, mother - Daughters, SD, mother 9.1 ( -0.4 to 18.7) 4.8 0.028 *
SD-5 Daughters, CyO, mother - Daughters, SD, mother -4.6 (-10.0 to 0.6) 2.7 0.041 *
SD-5 Daughters, SD, mother - Daughters, SD, father 5.9 ( -2.5 to 14.6) 4.2 0.07775
SD-5 Sons, CyO, father - Sons, SD, father -4.0 (-10.0 to 1.8) 3.0 0.08675
SD-5 Sons, SD, father - Daughters, SD, father 5.3 ( -3.7 to 14.5) 4.5 0.118
SD-5 Sons, CyO, mother - Sons, SD, mother -3.1 ( -9.0 to 2.8) 3.0 0.1493
SD-5 Daughters, CyO, mother - Daughters, CyO, father 2.5 ( -4.8 to 9.8) 3.7 0.2448
SD-5 Sons, CyO, father - Daughters, CyO, father 2.5 ( -4.9 to 10.4) 3.9 0.2597
SD-5 Daughters, CyO, father - Daughters, SD, father -1.2 ( -6.3 to 3.5) 2.5 0.3105
SD-72 Sons, SD, mother - Daughters, SD, mother 9.9 ( 4.2 to 15.8) 3.0 5e-04 *
SD-72 Sons, CyO, mother - Daughters, CyO, mother 11.0 ( 4.3 to 17.9) 3.5 0.00075 *
SD-72 Sons, CyO, father - Daughters, CyO, father 7.0 ( -1.0 to 15.5) 4.2 0.04475 *
SD-72 Sons, CyO, mother - Sons, SD, mother 3.4 ( -1.2 to 8.1) 2.4 0.074
SD-72 Daughters, SD, mother - Daughters, SD, father -3.8 ( -9.9 to 1.6) 2.9 0.08525
SD-72 Daughters, CyO, mother - Daughters, SD, mother 2.3 ( -1.1 to 5.9) 1.8 0.0875
SD-72 Sons, SD, father - Daughters, SD, father 4.5 ( -3.3 to 12.7) 4.1 0.1335
SD-72 Sons, CyO, father - Sons, SD, father 2.2 ( -3.8 to 8.8) 3.2 0.2433
SD-72 Sons, CyO, mother - Sons, CyO, father 2.7 ( -6.3 to 11.4) 4.5 0.2742
SD-72 Daughters, CyO, mother - Daughters, CyO, father -1.2 ( -7.4 to 4.7) 3.0 0.3395
SD-72 Sons, SD, mother - Sons, SD, father 1.5 ( -6.4 to 9.3) 4.1 0.354
SD-72 Daughters, CyO, father - Daughters, SD, father -0.3 ( -5.0 to 4.4) 2.4 0.449
SD-Mad Sons, CyO, mother - Daughters, CyO, mother 9.4 ( 2.7 to 16.3) 3.5 0.00225 *
SD-Mad Daughters, CyO, mother - Daughters, SD, mother -5.4 ( -9.9 to -1.2) 2.2 0.00675 *
SD-Mad Sons, CyO, mother - Sons, CyO, father 7.7 ( -0.6 to 15.6) 4.1 0.0345 *
SD-Mad Sons, CyO, mother - Sons, SD, mother 3.0 ( -1.9 to 8.0) 2.5 0.1115
SD-Mad Daughters, SD, mother - Daughters, SD, father 3.4 ( -4.9 to 11.0) 4.0 0.1965
SD-Mad Sons, CyO, father - Sons, SD, father -2.2 ( -8.7 to 4.1) 3.2 0.2318
SD-Mad Daughters, CyO, mother - Daughters, CyO, father -2.8 (-10.7 to 4.3) 3.8 0.236
SD-Mad Sons, SD, mother - Sons, SD, father 2.5 ( -6.3 to 10.3) 4.1 0.2578
SD-Mad Sons, SD, father - Daughters, SD, father 1.9 ( -6.7 to 10.9) 4.6 0.3347
SD-Mad Sons, SD, mother - Daughters, SD, mother 1.0 ( -6.1 to 8.0) 3.6 0.3847
SD-Mad Daughters, CyO, father - Daughters, SD, father 0.8 ( -5.2 to 6.9) 3.1 0.4085
SD-Mad Sons, CyO, father - Daughters, CyO, father -1.1 ( -9.8 to 7.8) 4.5 0.4127

sessionInfo()
R version 3.5.1 (2018-07-02)
Platform: x86_64-apple-darwin15.6.0 (64-bit)
Running under: macOS High Sierra 10.13.6

Matrix products: default
BLAS: /Library/Frameworks/R.framework/Versions/3.5/Resources/lib/libRblas.0.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/3.5/Resources/lib/libRlapack.dylib

locale:
[1] en_AU.UTF-8/en_AU.UTF-8/en_AU.UTF-8/C/en_AU.UTF-8/en_AU.UTF-8

attached base packages:
[1] grid      stats     graphics  grDevices utils     datasets  methods  
[8] base     

other attached packages:
 [1] gameofthrones_1.0.0 stringr_1.4.0       tibble_2.0.99.9000 
 [4] readr_1.1.1         ggthemes_4.0.1      gridExtra_2.3      
 [7] lazerhawk_0.2.4     pander_0.6.2        Cairo_1.5-9        
[10] reshape2_1.4.3      ggplot2_3.1.0       brms_2.8.0         
[13] Rcpp_1.0.1          dplyr_0.8.0.1       knitr_1.22         

loaded via a namespace (and not attached):
 [1] nlme_3.1-137         matrixStats_0.54.0   fs_1.3.1            
 [4] xts_0.11-0           devtools_1.13.6      RColorBrewer_1.1-2  
 [7] threejs_0.3.1        rprojroot_1.3-2      rstan_2.18.2        
[10] tools_3.5.1          backports_1.1.2      R6_2.4.0            
[13] DT_0.4               lazyeval_0.2.2       colorspace_1.3-2    
[16] withr_2.1.2          tidyselect_0.2.5     prettyunits_1.0.2   
[19] processx_3.2.1       Brobdingnag_1.2-5    compiler_3.5.1      
[22] git2r_0.23.0         cli_1.1.0            shinyjs_1.0         
[25] labeling_0.3         colourpicker_1.0     scales_1.0.0        
[28] dygraphs_1.1.1.6     mvtnorm_1.0-8        ggridges_0.5.0      
[31] callr_2.0.4          digest_0.6.18        StanHeaders_2.18.0  
[34] rmarkdown_1.10       base64enc_0.1-3      pkgconfig_2.0.2     
[37] htmltools_0.3.6      htmlwidgets_1.2      rlang_0.3.4         
[40] shiny_1.3.2          zoo_1.8-3            crosstalk_1.0.0     
[43] gtools_3.8.1         inline_0.3.15        magrittr_1.5        
[46] loo_2.1.0            bayesplot_1.6.0      Matrix_1.2-14       
[49] munsell_0.5.0        abind_1.4-5          stringi_1.4.3       
[52] whisker_0.3-2        yaml_2.2.0           MASS_7.3-50         
[55] pkgbuild_1.0.2       plyr_1.8.4           parallel_3.5.1      
[58] promises_1.0.1       crayon_1.3.4         miniUI_0.1.1.1      
[61] lattice_0.20-35      hms_0.4.2            ps_1.3.0            
[64] pillar_1.3.1.9000    igraph_1.2.1         markdown_0.9        
[67] shinystan_2.5.0      stats4_3.5.1         rstantools_1.5.0    
[70] glue_1.3.1.9000      evaluate_0.13        httpuv_1.5.1        
[73] tidyr_0.8.2          gtable_0.2.0         purrr_0.3.2         
[76] assertthat_0.2.1     xfun_0.6             mime_0.6            
[79] xtable_1.8-4         colortools_0.1.5     coda_0.19-2         
[82] later_0.8.0          rsconnect_0.8.8      shinythemes_1.1.1   
[85] memoise_1.1.0        workflowr_1.3.0      bridgesampling_0.4-0