1 CRF

1.0.1 Data import and formatting

M_CRF <- read.csv("M_CRF_combined.csv")
M_CRF <- standardise_levels(M_CRF)

1.1 Global model.

glmer((target offspring, competitor offspring) ~ Selection x Female x Background x Replicate + (1|Contest), family = binomial).

Overall, ML males displayed greater mating success, with a model-estimated probability of siring offspring of 32.5%, relative to 24.3% for MC males. However, this effect also depended on interactions with female type and male genetic background, as detailed below.

m_crf <- glmer(
  cbind(Red, U) ~
    Selection*Female*Background*Replicate +
    (1|Contest),
  data = na.omit(M_CRF),
  family = "binomial"
)

rep_anova(
  car::Anova(m_crf),
  title = "Table S1.1: Competitive reproductive fitness"
) %>% tab_options(table.font.size = px(13), data_row.padding = px(4))

Term	Chisq	Df	Pr(>Chisq)
Table S1.1: Competitive reproductive fitness
Selection	56.448	1	< 0.001
Female	0.436	1	0.509
Background	241.299	1	< 0.001
Replicate	1.785	2	0.410
Selection:Female	3.774	1	0.052
Selection:Background	17.446	1	< 0.001
Female:Background	49.498	1	< 0.001
Selection:Replicate	1.170	2	0.557
Female:Replicate	4.302	2	0.116
Background:Replicate	0.477	2	0.788
Selection:Female:Background	0.171	1	0.679
Selection:Female:Replicate	0.020	2	0.990
Selection:Background:Replicate	3.459	2	0.177
Female:Background:Replicate	0.844	2	0.656
Selection:Female:Background:Replicate	5.755	2	0.056

1.2 Contrast testing.

Multiple-testing correction. Adjusted alpha for 4 contrast tests: ~0.013.

alpha_adj <- 1 - (0.95)^(1/4)

em <- emmeans(m_crf,
  ~ Selection | Female * Background)

contrasts <- pairs(em) %>% as.data.frame()

rep_contrasts(contrasts, alpha_adj,
  "Table S1.2: Male CRF Contrasts") %>% 
  tab_options(table.font.size = px(13), data_row.padding = px(4))

contrast	Female	Background	estimate	SE	df	z.ratio	p.value
Table S1.2: Male CRF Contrasts
ML - MC	CG	CG	0.852	0.133	Inf	6.414	< 0.001
ML - MC	C	CG	0.661	0.137	Inf	4.831	< 0.001
ML - MC	CG	C	0.370	0.114	Inf	3.254	0.001
ML - MC	C	C	0.059	0.139	Inf	0.422	0.673

1.3 Dataset visualisation

plotDat <- make_emm(m_crf, response = T)
contrasts_df <- sig_ext(contrasts)
plot_glm_means_ci(plotDat, contrasts_df, title="Competitive reproductive fitness") +
  labs(
    x = "Genetic background",
    y = "Proportion sired",
    colour = "Selection"
  )

1.4 Assumption checks

Show assumption checks

sim_res <- simulateResiduals(m_crf)

plot(sim_res)

par(mfrow = c(2,2))

testDispersion(sim_res)

## 
##  DHARMa nonparametric dispersion test via sd of residuals fitted vs.
##  simulated
## 
## data:  simulationOutput
## dispersion = 0.97573, p-value = 0.72
## alternative hypothesis: two.sided

testOutliers(sim_res)

## 
##  DHARMa outlier test based on exact binomial test with approximate
##  expectations
## 
## data:  sim_res
## outliers at both margin(s) = 5, observations = 666, p-value = 1
## alternative hypothesis: true probability of success is not equal to 0.007968127
## 95 percent confidence interval:
##  0.00244203 0.01743245
## sample estimates:
## frequency of outliers (expected: 0.00796812749003984 ) 
##                                            0.007507508

testUniformity(sim_res)

## 
##  Asymptotic one-sample Kolmogorov-Smirnov test
## 
## data:  simulationOutput$scaledResiduals
## D = 0.04634, p-value = 0.1145
## alternative hypothesis: two-sided

par(mfrow = c(1,1))

2 MC success

2.0.1 Data import and formatting

M_MC <- read.csv("M_MC_combined.csv")
M_MC <- standardise_levels(M_MC)
M_MC$Success <- as.integer(M_MC$Mated.male == "wt")

2.1 Global model.

glm(Success ~ Selection x Female x Background x Replicate, family = binomial).

Overall, ML males were more successful in mate-choice trials, winning 305 of 535 trials, compared with 246 of 528 trials for MC males.

m_mc <- glm(
  Success ~
    Selection*Female*Background*Replicate,
  data = M_MC,
  family = "binomial"
)

rep_anova(
  car::Anova(m_mc),
  title = "Table S2.1: Mating success"
) %>% tab_options(table.font.size = px(13), data_row.padding = px(4))

Term	LR Chisq	Df	Pr(>Chisq)
Table S2.1: Mating success
Selection	12.114	1	0.001
Female	6.145	1	0.013
Background	1.928	1	0.165
Replicate	14.171	2	0.001
Selection:Female	0.212	1	0.645
Selection:Background	0.035	1	0.852
Female:Background	0.100	1	0.752
Selection:Replicate	0.591	2	0.744
Female:Replicate	0.005	2	0.997
Background:Replicate	1.326	2	0.515
Selection:Female:Background	0.414	1	0.520
Selection:Female:Replicate	5.153	2	0.076
Selection:Background:Replicate	2.531	2	0.282
Female:Background:Replicate	1.481	2	0.477
Selection:Female:Background:Replicate	0.972	2	0.615

2.2 Contrast testing.

No post hoc contrasts were required for this analysis.

alpha_adj <- 1 - (0.95)^(1/12)

em <- emmeans(m_mc,
  ~ Selection | Female * Background * Replicate)

contrasts <- pairs(em) %>% as.data.frame()

#rep_contrasts(contrasts, alpha_adj, "Table S2.2: Male mating success contrasts") %>% 
  #tab_options(table.font.size = px(13), #data_row.padding = px(4))

2.3 Dataset visualisation

plotDat <- make_emm(m_mc, response = T)
contrasts_df <- sig_ext(contrasts)
plot_glm_means_ci(plotDat, contrasts_df, title="Mating success") + 
  labs(
    x = "Genetic background",
    y = "Proportion of successful trials",
    colour = "Selection"
  )

2.4 Assumption checks

Show assumption checks

sim_res <- simulateResiduals(m_mc)

plot(sim_res)

par(mfrow = c(2,2))

testDispersion(sim_res)

## 
##  DHARMa nonparametric dispersion test via sd of residuals fitted vs.
##  simulated
## 
## data:  simulationOutput
## dispersion = 1.0001, p-value = 0.992
## alternative hypothesis: two.sided

testOutliers(sim_res)

## 
##  DHARMa outlier test based on exact binomial test with approximate
##  expectations
## 
## data:  sim_res
## outliers at both margin(s) = 14, observations = 1108, p-value = 0.08833
## alternative hypothesis: true probability of success is not equal to 0.007968127
## 95 percent confidence interval:
##  0.006924627 0.021109226
## sample estimates:
## frequency of outliers (expected: 0.00796812749003984 ) 
##                                             0.01263538

testUniformity(sim_res)

## 
##  Asymptotic one-sample Kolmogorov-Smirnov test
## 
## data:  simulationOutput$scaledResiduals
## D = 0.024123, p-value = 0.5393
## alternative hypothesis: two-sided

3 MC latency

3.0.1 Data import and filtering

M_MC <- M_MC %>% filter(Mated.male == "wt",
    (Mating.end.time - Mating.start.time) > 0,
    (Mating.start.time - Obs.start.time) > 0)

3.1 Global model

lm(log(latency + 1) ~ Selection x Female x Background x Replicate)

Successful ML males displayed lower latency to mating than MC males, mating approximately 18% faster on average (14.1 min vs 17.2 min). This likely reflects the greater mating success observed above

m_ml <- lm(
  log((Mating.start.time - Obs.start.time) + 1) ~
    Selection*Female*Background*Replicate,
  data = na.omit(M_MC)
)

rep_anova(
  car::Anova(m_ml),
  title = "Table S3: Mating latency"
) %>% tab_options(table.font.size = px(13), data_row.padding = px(4))

Term	Sum Sq	Df	F value	Pr(>F)
Table S3: Mating latency
Selection	2.725	1	3.907	0.049
Female	0.149	1	0.214	0.644
Background	0.191	1	0.274	0.601
Replicate	2.937	2	2.106	0.123
Selection:Female	0.048	1	0.069	0.792
Selection:Background	0.136	1	0.195	0.659
Female:Background	1.447	1	2.075	0.150
Selection:Replicate	2.453	2	1.759	0.173
Female:Replicate	1.104	2	0.791	0.454
Background:Replicate	0.163	2	0.117	0.890
Selection:Female:Background	0.055	1	0.078	0.780
Selection:Female:Replicate	2.732	2	1.959	0.142
Selection:Background:Replicate	0.155	2	0.111	0.895
Female:Background:Replicate	0.876	2	0.628	0.534
Selection:Female:Background:Replicate	2.028	2	1.454	0.235
Residuals	363.338	521

3.2 Contrast testing

No post hoc contrasts were required for this analysis.

alpha_adj <- 1 - (0.95)^(1)

3.3 Dataset visualisation

plotDat <- make_emm(m_ml, response = F)
plot_lm_means_ci(plotDat, NULL, title="Mating latency") + 
  labs(
    x = "Genetic background",
    y = "log(latency+1)",
    colour = "Selection"
  )

3.4 Assumption checks

Show assumption checks

par(mfrow = c(2,2))
plot(m_ml)

par(mfrow = c(1,1))

Residual diagnostics. Residuals deviate modestly from normality; however, the deviations are minor and the model is expected to remain robust to them.

hist(
  resid(m_ml),
  main = "Residual distribution",
  xlab = "Residuals"
)

shapiro.test(resid(m_ml))

## 
##  Shapiro-Wilk normality test
## 
## data:  resid(m_ml)
## W = 0.97227, p-value = 1.228e-08

4 MC duration

4.0.1 Data import and filtering

# already covered above

4.1 Global model

lm(Duration ~ Selection x Female x Background x Replicate)

MC males mated for longer durations on average, mating approximately 6% longer than ML males (20.6 min vs 19.2 min). If mating duration reflects sperm investment or transfer, this suggests that ML males are not competitive through prolonged copulation. Alternatively, shorter mating durations may reflect the evolution of reduced mate harm.

m_md <- lm(
  I(Mating.end.time - Mating.start.time) ~
    Selection*Female*Background*Replicate,
  data = na.omit(M_MC)
)

rep_anova(
  car::Anova(m_md),
  title = "Table S4: Mating duration"
) %>% tab_options(table.font.size = px(13), data_row.padding = px(4))

Term	Sum Sq	Df	F value	Pr(>F)
Table S4: Mating duration
Selection	217.422	1	6.953	0.009
Female	223.731	1	7.155	0.008
Background	243.567	1	7.789	0.005
Replicate	178.077	2	2.847	0.059
Selection:Female	0.755	1	0.024	0.877
Selection:Background	55.051	1	1.760	0.185
Female:Background	1.743	1	0.056	0.813
Selection:Replicate	79.312	2	1.268	0.282
Female:Replicate	20.808	2	0.333	0.717
Background:Replicate	38.351	2	0.613	0.542
Selection:Female:Background	46.347	1	1.482	0.224
Selection:Female:Replicate	2.794	2	0.045	0.956
Selection:Background:Replicate	7.923	2	0.127	0.881
Female:Background:Replicate	34.590	2	0.553	0.576
Selection:Female:Background:Replicate	84.408	2	1.350	0.260
Residuals	16,291.969	521

4.2 Contrast testing

No post hoc contrasts were required for this analysis.

alpha_adj <- 1 - (0.95)^(1/4)

em <- emmeans(m_md,
  ~ Selection | Female * Background)

contrasts <- pairs(em) %>% as.data.frame()

4.3 Dataset visualisation

plotDat <- make_emm(m_md, response = F)
contrasts_df <- sig_ext(contrasts)
plot_lm_means_ci(plotDat, NULL, title="Mating duration") + 
  labs(
    x = "Genetic background",
    y = "Mating Duration (mins)",
    colour = "Selection"
  )

4.4 Assumption checks

Show assumption checks

par(mfrow = c(2,2))
plot(m_md)

par(mfrow = c(1,1))

Residual diagnostics:

Residuals were broadly normally distributed, with only a small number of apparent outliers. The model is expected to be robust to these deviations.

hist(
  resid(m_md),
  main = "Residual distribution",
  xlab = "Residuals"
)

shapiro.test(resid(m_md))

## 
##  Shapiro-Wilk normality test
## 
## data:  resid(m_md)
## W = 0.87925, p-value < 2.2e-16

5 MC fecundity induction

5.0.1 Data import and formatting

M_MC <- M_MC %>% mutate(fec = Offspring_f + Offspring_m)

5.1 Global model

lm(fecundity ~ Selection x Female x Background x Replicate)

No detectable differences were observed between ML and MC males in the number of offspring sired following a successful mating.

m_fi <- lm(
  fec ~
    Selection*Female*Background*Replicate,
  data = na.omit(M_MC)
)

rep_anova(
  car::Anova(m_fi),
  title = "Table S5: Offspring sired"
) %>% tab_options(table.font.size = px(13), data_row.padding = px(4))

Term	Sum Sq	Df	F value	Pr(>F)
Table S5: Offspring sired
Selection	110.939	1	0.777	0.379
Female	107,596.815	1	753.372	< 0.001
Background	22,515.817	1	157.651	< 0.001
Replicate	1,106.049	2	3.872	0.021
Selection:Female	10.602	1	0.074	0.785
Selection:Background	262.289	1	1.836	0.176
Female:Background	3,069.929	1	21.495	< 0.001
Selection:Replicate	81.396	2	0.285	0.752
Female:Replicate	416.910	2	1.460	0.233
Background:Replicate	67.194	2	0.235	0.790
Selection:Female:Background	13.843	1	0.097	0.756
Selection:Female:Replicate	1.242	2	0.004	0.996
Selection:Background:Replicate	427.150	2	1.495	0.225
Female:Background:Replicate	62.697	2	0.219	0.803
Selection:Female:Background:Replicate	443.010	2	1.551	0.213
Residuals	74,409.382	521

5.2 Contrast testing

No post hoc contrasts were required for this analysis.

alpha_adj <- 1 - (0.95)^(1)

5.3 Dataset visualisation

plotDat <- make_emm(m_fi, response = F)
plot_lm_means_ci(plotDat, NULL, title="Offspring sired") + 
  labs(
    x = "Genetic background",
    y = "Fecundity Induced",
    colour = "Selection"
  ) + scale_y_continuous(limits = c(0, 65))

5.4 Assumption checks

Show assumption checks

par(mfrow = c(2,2))
plot(m_fi)

par(mfrow = c(1,1))

Residuals revealed minor deviations from normality.

hist(
  resid(m_fi),
  main = "Residual distribution",
  xlab = "Residuals"
)

shapiro.test(resid(m_fi))

## 
##  Shapiro-Wilk normality test
## 
## data:  resid(m_fi)
## W = 0.9313, p-value = 4.067e-15

6 MC Brood sex ratios

6.0.1 Data import and formatting

M_MC <- M_MC %>%
  filter(Offspring_m + Offspring_f != 0)

6.1 Global model

lm(fecundity ~ Selection x Female x Background x Replicate)

No detectable differences were observed between ML and MC males in the number of offspring sired following a successful mating.

m_bs <- glm(
  cbind(Offspring_m, Offspring_f) ~
    Selection*Female*Background*Replicate,
  data = na.omit(M_MC), 
  family = "binomial"
)

rep_anova(
  car::Anova(m_bs),
  title = "Table S6: Brood sex ratio"
) %>% tab_options(table.font.size = px(13), data_row.padding = px(4))

Term	LR Chisq	Df	Pr(>Chisq)
Table S6: Brood sex ratio
Selection	1.955	1	0.162
Female	0.084	1	0.771
Background	0.391	1	0.532
Replicate	0.339	2	0.844
Selection:Female	2.434	1	0.119
Selection:Background	0.357	1	0.550
Female:Background	0.099	1	0.752
Selection:Replicate	5.648	2	0.059
Female:Replicate	2.306	2	0.316
Background:Replicate	4.772	2	0.092
Selection:Female:Background	0.598	1	0.439
Selection:Female:Replicate	3.564	2	0.168
Selection:Background:Replicate	0.796	2	0.672
Female:Background:Replicate	0.575	2	0.750
Selection:Female:Background:Replicate	0.639	2	0.727

6.2 Contrast testing

No post hoc contrasts were required for this analysis.

alpha_adj <- 1 - (0.95)^(1)
em <- emmeans(m_bs,
  ~ Selection | Female * Background)
contrasts <- pairs(em) %>% as.data.frame()
contrasts_df <- sig_ext(contrasts)

6.3 Dataset visualisation

plotDat <- make_emm(m_bs, response = T)
plot_glm_means_ci(plotDat, contrasts_df, title="Brood sex ratio")  + 
  labs(
    x = "Genetic background",
    y = "Sex ratio (M/F)",
    colour = "Selection"
  )

6.4 Assumption checks

Show assumption checks

sim_res <- simulateResiduals(m_bs)

plot(sim_res)

par(mfrow = c(2,2))

testDispersion(sim_res)

## 
##  DHARMa nonparametric dispersion test via sd of residuals fitted vs.
##  simulated
## 
## data:  simulationOutput
## dispersion = 0.98237, p-value = 0.504
## alternative hypothesis: two.sided

testOutliers(sim_res)

## 
##  DHARMa outlier test based on exact binomial test with approximate
##  expectations
## 
## data:  sim_res
## outliers at both margin(s) = 8, observations = 515, p-value = 0.07304
## alternative hypothesis: true probability of success is not equal to 0.007968127
## 95 percent confidence interval:
##  0.006729707 0.030377776
## sample estimates:
## frequency of outliers (expected: 0.00796812749003984 ) 
##                                             0.01553398

testUniformity(sim_res)

## 
##  Asymptotic one-sample Kolmogorov-Smirnov test
## 
## data:  simulationOutput$scaledResiduals
## D = 0.040879, p-value = 0.3557
## alternative hypothesis: two-sided

7 SC sperm offense

7.0.1 Data import and formatting

M_SC <- read.csv("M_SC_combined.csv")
M_SC <- standardise_levels(M_SC)

7.1 Global model

glmmTMB(cbind(Competitor offspring, Target offspring) ~ Selection x Female x Background x Replicate + (1|Contest), family = betabinomial)

This analysis models sperm defense (1 - P2 -> modified to solve an overdispersion issue), representing the probability that the first (competitor) male successfully defended paternity against the focal second male (ML or MC). No detectable differences were observed between ML and MC males in sperm offense ability. Model-estimated probabilities of successful second-male paternity were 83.4% for ML males and 81.6% for MC males.

m_sc <- glmmTMB(
  cbind(U, WT) ~
    Selection*Female*Background*Replicate +
    (1|Contest),
  data = M_SC,
  family = "betabinomial"
)

rep_anova(
  car::Anova(m_sc),
  title = "Table S7: Sperm offense"
) %>% tab_options(table.font.size = px(13), data_row.padding = px(4))

Term	Chisq	Df	Pr(>Chisq)
Table S7: Sperm offense
Selection	1.573	1	0.210
Female	115.819	1	< 0.001
Background	69.423	1	< 0.001
Replicate	1.937	2	0.380
Selection:Female	0.001	1	0.974
Selection:Background	< 0.001	1	0.997
Female:Background	74.341	1	< 0.001
Selection:Replicate	2.017	2	0.365
Female:Replicate	1.846	2	0.397
Background:Replicate	0.953	2	0.621
Selection:Female:Background	0.750	1	0.387
Selection:Female:Replicate	0.567	2	0.753
Selection:Background:Replicate	5.094	2	0.078
Female:Background:Replicate	0.468	2	0.791
Selection:Female:Background:Replicate	5.029	2	0.081

7.2 Contrast testing

No post hoc contrasts were required for this analysis.

alpha_adj <- 1 - (0.95)^(1/4)

em <- emmeans(m_sc,
  ~ Selection | Female * Background * Replicate)

contrasts <- pairs(em) %>% as.data.frame()
contrasts_df <- sig_ext(contrasts)

7.3 Dataset visualisation

plotDat <- make_emm(m_sc, response = T)
plot_glm_means_ci(plotDat, contrasts_df, title="Sperm offense") + 
  labs(
    x = "Genetic background",
    y = "Competitor Sperm Defense (1-P2)",
    colour = "Selection"
  )

7.4 Assumption checks

Show assumption checks

sim_res <- simulateResiduals(m_sc)

plot(sim_res)

par(mfrow = c(2,2))

testDispersion(sim_res)

## 
##  DHARMa nonparametric dispersion test via sd of residuals fitted vs.
##  simulated
## 
## data:  simulationOutput
## dispersion = 0.83713, p-value = 0.152
## alternative hypothesis: two.sided

testOutliers(sim_res)

## 
##  DHARMa outlier test based on exact binomial test with approximate
##  expectations
## 
## data:  sim_res
## outliers at both margin(s) = 16, observations = 1152, p-value = 0.0305
## alternative hypothesis: true probability of success is not equal to 0.007968127
## 95 percent confidence interval:
##  0.007958981 0.022456923
## sample estimates:
## frequency of outliers (expected: 0.00796812749003984 ) 
##                                             0.01388889

testUniformity(sim_res)

## 
##  Asymptotic one-sample Kolmogorov-Smirnov test
## 
## data:  simulationOutput$scaledResiduals
## D = 0.015765, p-value = 0.937
## alternative hypothesis: two-sided

8 MH CG Mate Harm

8.0.1 Data import and formatting

M_MH <- read.csv("M_MH_CG.csv")
M_MH <- standardise_levels(M_MH)

MH <- M_MH[,-5] %>%
  pivot_longer(
    cols = starts_with("Day"),
    names_to = "Day",
    names_prefix = "Day",
    values_to = "alive"
  ) %>%
  mutate(Day = as.numeric(Day)) %>%
  group_by(Contest) %>%
  mutate(
    at_risk = lag(alive),
    deaths = at_risk - alive
  ) %>% ungroup()

8.1 Global model

glmer((deaths, survival) ~ Selection x Background x Replicate x Day + (1|Contest), family = binomial)

CG Female survivorship across seven days (under continuous exposure to males during days 0–2, followed by male removal) was modelled using a discrete-time survival model. Females housed with ML males displayed consistently higher survivorship than those housed with MC males. Survivorship declined more sharply during the cohabitation phase in both treatments. Final survival probabilities were estimated at approximately 59% for ML treatments and 45% for MC treatments.

m_mh <- glmer(
  cbind(deaths, at_risk - deaths) ~ 
    Selection * Background * Replicate * factor(Day) + (1|Contest),
  data = MH, 
  family = binomial)

rep_anova(
  car::Anova(m_mh),
  title = "Table S8: Mate harm / Clone Generator female survival"
) %>% tab_options(table.font.size = px(13), data_row.padding = px(4))

Term	Chisq	Df	Pr(>Chisq)
Table S8: Mate harm / Clone Generator female survival
Selection	22.910	8	0.003
Background	22.033	6	0.001
Replicate	8.681	9	0.467
factor(Day)	332.301	18	< 0.001
Selection:Background	3.099	4	0.541
Selection:Replicate	2.722	4	0.605
Background:Replicate	5.996	4	0.199
Selection:factor(Day)	7.546	11	0.753
Background:factor(Day)	10.936	10	0.363
Replicate:factor(Day)	9.966	16	0.868
Selection:Background:Replicate	0.980	2	0.613
Selection:Background:factor(Day)	1.364	6	0.968
Selection:Replicate:factor(Day)	9.410	12	0.668
Background:Replicate:factor(Day)	13.290	12	0.348
Selection:Background:Replicate:factor(Day)	8.226	12	0.767

8.2 Contrast testing

No post hoc contrasts were required for this analysis.

alpha_adj <- 1 - (0.95)^(1/2)

em <- emmeans(m_mh,  ~ Selection | Background * factor(Day))
contrasts <- pairs(em) %>% as.data.frame()

#rep_contrasts(contrasts, alpha_adj, 
#              "Table S8.2: Mate Harm") %>% #tab_options(table.font.size = px(13), data_row.padding = px(4))

8.3 Dataset visualisation

plotDat <- emmeans(m_mh,
  ~ Selection * Background * Day,
  type = "response") %>%
  as.data.frame() %>%
  arrange(Selection, Background, Day) %>%
  group_by(Selection, Background) %>%
  mutate(
    survival = cumprod(1 - prob)
  )
d0 <- plotDat %>% distinct(Selection, Background) %>%
  mutate(Day = 0, survival = 1); plotDat <- bind_rows(d0, plotDat)


contrasts_df <- sig_ext(contrasts) 

ggplot(plotDat, aes(x = Day, y = survival, colour = Selection)) +
  geom_line() +
  geom_text(data = contrasts_df,
    aes(x = 3.5, y = 0.95, label = sig), inherit.aes = FALSE, size = 6) +
facet_wrap(~ Background, labeller = as_labeller(
      function(x) paste("Genetic background:", x))) +
  
  labs(title = "Mate harm / CG female survival",
    x = "Time", y = "Proportion of survivors",
    colour = "Selection"
  ) +
  
  scale_y_continuous(limits = c(0, 1)) + theme_classic(base_size = 14) +
  
  theme(plot.title = element_text(
      face = "bold", hjust = 0.5, size = 16),
    strip.background = element_blank(),
    strip.text = element_text(face = "bold", size = 12),
    axis.title = element_text(face = "bold"),
    legend.position = "right", legend.title = element_text(face = "bold"),
    panel.spacing = unit(1, "lines")
  )

8.4 Assumption checks

Show assumption checks

sim_res <- simulateResiduals(m_mh)

plot(sim_res)

par(mfrow = c(2,2))

testDispersion(sim_res)

## 
##  DHARMa nonparametric dispersion test via sd of residuals fitted vs.
##  simulated
## 
## data:  simulationOutput
## dispersion = 0.91996, p-value = 0.36
## alternative hypothesis: two.sided

testOutliers(sim_res)

## 
##  DHARMa outlier test based on exact binomial test with approximate
##  expectations
## 
## data:  sim_res
## outliers at both margin(s) = 12, observations = 840, p-value = 0.04938
## alternative hypothesis: true probability of success is not equal to 0.007968127
## 95 percent confidence interval:
##  0.007402845 0.024821187
## sample estimates:
## frequency of outliers (expected: 0.00796812749003984 ) 
##                                             0.01428571

testUniformity(sim_res)

## 
##  Asymptotic one-sample Kolmogorov-Smirnov test
## 
## data:  simulationOutput$scaledResiduals
## D = 0.024811, p-value = 0.6793
## alternative hypothesis: two-sided

9 MH CG female fecundity

9.0.1 Data import and formatting

# no need for input

9.1 Global model

lm(Productivity ~ Selection x Background x Replicate)

In addition to improving female survivorship, ML males were also associated with greater offspring production. However, this effect depended on the genetic background in which males were tested, with the productivity benefit observed only under the WT background.

m_pr <- lm(
  Productivity ~
    Selection*Background*Replicate,
  data = na.omit(M_MH)
)

rep_anova(
  car::Anova(m_pr),
  title = "Table S9.1: CG female productivity"
) %>% tab_options(table.font.size = px(13), data_row.padding = px(4))

Term	Sum Sq	Df	F value	Pr(>F)
Table S9.1: CG female productivity
Selection	7,760.208	1	21.924	< 0.001
Background	58,124.008	1	164.209	< 0.001
Replicate	430.317	2	0.608	0.546
Selection:Background	11,505.208	1	32.504	< 0.001
Selection:Replicate	63.517	2	0.090	0.914
Background:Replicate	363.217	2	0.513	0.600
Selection:Background:Replicate	1,171.217	2	1.654	0.196
Residuals	38,228.100	108

9.2 Contrast testing

Multiple-testing correction. Adjusted alpha for 2 contrast tests: ~0.025

alpha_adj <- 1 - (0.95)^(1/2)

em <- emmeans(m_pr,  ~ Selection | Background)
contrasts <- pairs(em) %>% as.data.frame()

rep_contrasts(contrasts, alpha_adj, 
              "Table S9.2: CG female productivity") %>% 
  tab_options(table.font.size = px(13), data_row.padding = px(4))

contrast	Background	estimate	SE	df	t.ratio	p.value
Table S9.2: CG female productivity
ML - MC	CG	−3.500	4.858	108	−0.721	0.473
ML - MC	C	35.667	4.858	108	7.342	< 0.001

9.3 Dataset visualisation

plotDat <- as.data.frame(emmeans(m_pr, ~ Selection * Background))
contrasts_df <- sig_ext(contrasts)

ggplot(plotDat, aes(x = Background, y = emmean, colour = Selection)) +
    geom_point(position = position_dodge(width = 0.25), size = 2.2) +
    geom_errorbar(aes(ymin = lower.CL, ymax = upper.CL),
                  width = 0.15, linewidth = 0.7,
                  position = position_dodge(width = 0.25)) +
  
    geom_text(data = contrasts_df, aes(x = Background, y = 125, label = sig),
      inherit.aes = FALSE, size = 5) +

    labs(title = "CG female productivity", 
         x = "Genetic background", y = "Estimated mean ± 95% CI", 
         colour = "Selection") +
    scale_y_continuous(limits = c(0,150), expand = expansion(mult = c(0, 0.02))) +
    
    theme_classic(base_size = 14) +
    theme(plot.title = element_text(face = "bold", hjust = 0.5, size = 16),
      axis.title = element_text(face = "bold"),
      legend.position = "right",
      legend.title = element_text(face = "bold"))

9.4 Assumption checks

Show assumption checks

par(mfrow = c(2,2))
plot(m_pr)

par(mfrow = c(1,1))

hist(
  resid(m_pr),
  main = "Residual distribution",
  xlab = "Residuals"
)

shapiro.test(resid(m_pr))

## 
##  Shapiro-Wilk normality test
## 
## data:  resid(m_pr)
## W = 0.99696, p-value = 0.9971

10 Isolated CG genetic effects

10.0.1 Data import and formatting

CG <- list(
  f_Auto = read.csv("ArtF_autoCRF.csv"),
  m_Auto = read.csv("ArtM_autoCRF.csv"),
  f_Cyto = read.csv("ArtF_cytoCRF.csv"),
  m_Cyto = read.csv("ArtM_cytoCRF.csv"),
  m_Y    = read.csv("ArtM_yCRF.csv")
)

cols <- c("Replicate", "Contest", "Cyto", "Auto", "Y")
CG <- map(CG, \(df) {
  df[cols] <- map(df[cols], factor)
  df$Selection <- factor(df$Selection, levels = c("ML", "MC"))
  df})

10.1 Global models

glmer((target offspring, competitor offspring) ~ Selection x Replicate + (1|Contest), family = binomial).

Given that in the absence of artefacts, the fitness difference between ML and MC males is marginal, any ML advantage in the presence of a given artefact suggests either compensatory adaptation to its imposed cost or epistatic coupling between the adaptive response and said genetic element. Likewise, in the absence of CG elements, ML females exhibit lower fitness than MC females. Persistence of this pattern in the presence of a ML-adapted CG element (MC > ML female fitness) suggests sex-specific or sexually antagonistic effects, whereas recovery of ML female fitness indicates concordant fitness benefits across sexes

We find the following:

CG genetic element	ML male response	ML female response
CG cytoplasm	Fitness benefit	Fitness recovery
CG autosomes	No fitness benefit	No fitness recovery
CG Y chromosome	Fitness benefit	Not applicable

cg <- map(CG, \(dat) {
  glmer(
    cbind(WT, U) ~ Selection * Replicate + (1 | Contest),
    data = dat,
    family = binomial
  )
})

anova_list <- imap(cg, \(mod, nm) {
  car::Anova(mod, type = 2) %>%
    as.data.frame() %>%
    rownames_to_column("term") %>%
    mutate(dataset = nm)
})

anova_df <- bind_rows(anova_list) %>%
  mutate(model = case_when(
      dataset == "f_Auto" ~ "S10.1a: CG autosomes in females",
      dataset == "m_Auto" ~ "S10.1b: CG autosomes in males",
      dataset == "f_Cyto" ~ "S10.1c: CG cytoplasm in females",
      dataset == "m_Cyto" ~ "S10.1d: CG cytoplasm in males",
      dataset == "m_Y"    ~ "S10.1e: CG Y chromosome in males"),
    
    sig = `Pr(>Chisq)` < 0.05,
    Chisq = ifelse(Chisq < 0.001,"< 0.001", sprintf("%.3f", round(Chisq, 3))),
    `Pr(>Chisq)` = ifelse(`Pr(>Chisq)` < 0.001,"< 0.001", 
                          sprintf("%.3f", round(`Pr(>Chisq)`, 3)))) %>%
    select(model,term,Chisq,Df,`Pr(>Chisq)`,sig)

anova_df %>%
  gt(groupname_col = "model",rowname_col = "term") %>%
  cols_label(Chisq = "Chi-sq", Df = "Df", `Pr(>Chisq)` = "Pr(>Chisq)") %>%
  cols_hide(columns = sig) %>% tab_style(style = list(cell_fill(color = "#DFF0D8")),
  locations = cells_body(rows = sig == TRUE)) %>%
  tab_options(table.font.size = px(13), data_row.padding = px(4))

	Chi-sq	Df	Pr(>Chisq)
S10.1a: CG autosomes in females
Selection	31.838	1	< 0.001
Replicate	12.613	2	0.002
Selection:Replicate	20.751	2	< 0.001
S10.1b: CG autosomes in males
Selection	2.608	1	0.106
Replicate	0.706	2	0.702
Selection:Replicate	1.215	2	0.545
S10.1c: CG cytoplasm in females
Selection	0.227	1	0.634
Replicate	9.217	2	0.010
Selection:Replicate	10.280	2	0.006
S10.1d: CG cytoplasm in males
Selection	5.672	1	0.017
Replicate	0.039	2	0.981
Selection:Replicate	5.943	2	0.051
S10.1e: CG Y chromosome in males
Selection	36.657	1	< 0.001
Replicate	31.006	2	< 0.001
Selection:Replicate	12.860	2	0.002

10.2 Contrast testing

Contrast testing conducted for 3 models with Selection:Replicate interactions.

Multiple-testing correction: Adjusted alpha for 3 contrast tests: ~0.017

alpha_adj <- 1 - (0.95)^(1/3)

cg_sub <- cg[c("f_Auto", "f_Cyto", "m_Y")]

contrast_df <- imap_dfr(cg_sub, \(mod, nm) {
  em <- emmeans(mod, ~ Selection | Replicate)
  pairs(em) %>% 
    as.data.frame() %>%
    mutate(dataset = nm)
})

contrast_gt_df <- contrast_df %>%
  mutate(
    model = case_when(
      dataset == "f_Auto" ~ "S10.2a: CG autosomes in females",
      dataset == "f_Cyto" ~ "S10.2b: CG cytoplasm in females",
      dataset == "m_Y"    ~ "S10.2c: CG Y chromosome in males"),
    
    sig = p.value < alpha_adj,
    
    estimate = ifelse(abs(estimate) < 0.001, "< 0.001",
      sprintf("%.3f", round(estimate, 3))),
    
    SE = ifelse(abs(SE) < 0.001,"< 0.001",
      sprintf("%.3f", round(SE, 3))),
    
    z.ratio = ifelse(abs(z.ratio) < 0.001,"< 0.001",
      sprintf("%.3f", round(z.ratio, 3))),
    
    p.value = ifelse(p.value < 0.001,"< 0.001",
      sprintf("%.3f", round(p.value, 3)))) %>%
  
  select(model,Replicate,contrast,estimate,SE,z.ratio,p.value,sig)

contrast_gt_df %>%
  gt(groupname_col = "model",rowname_col = "Replicate") %>%
  cols_label(contrast = "Contrast",estimate = "Estimate",SE = "SE",z.ratio = "z",p.value = "p") %>%
  cols_hide(columns = sig) %>%
  tab_style(style = list(cell_fill(color = "#DFF0D8")),
    locations = cells_body(rows = sig == TRUE))

	Contrast	Estimate	SE	z	p
S10.2a: CG autosomes in females
1	ML - MC	0.009	0.092	0.101	0.919
3	ML - MC	-0.327	0.093	-3.520	< 0.001
5	ML - MC	-0.579	0.091	-6.339	< 0.001
S10.2b: CG cytoplasm in females
1	ML - MC	0.130	0.102	1.276	0.202
3	ML - MC	0.084	0.092	0.907	0.364
5	ML - MC	-0.259	0.091	-2.838	0.005
S10.2c: CG Y chromosome in males
1	ML - MC	0.528	0.124	4.253	< 0.001
3	ML - MC	0.730	0.131	5.572	< 0.001
5	ML - MC	0.081	0.131	0.618	0.537

10.3 Dataset visualisation

emm_df <- imap_dfr(cg, \(mod, nm) {
  em <- emmeans(mod, ~ Selection, type = "response")
  as.data.frame(em) %>% mutate(dataset = nm)
})

contrast_df <- imap_dfr(cg, \(mod, nm) {
  em <- emmeans(mod, ~ Selection, type = "response")
  pairs(em) %>% as.data.frame() %>%
    mutate(
      dataset = nm,
      sig = case_when(
        p.value < 0.001 ~ "***",
        p.value < 0.01  ~ "**",
        p.value < 0.05  ~ "*",
        TRUE ~ ""
      )
    )
})

facet_labs <- c(
  f_Auto = "CG autosomes: ♀",
  m_Auto = "CG autosomes: ♂",
  f_Cyto = "CG cytoplasm: ♀",
  m_Cyto = "CG cytoplasm: ♂",
  m_Y    = "CG Y chr: ♂"
)

ggplot(emm_df, aes(x = Selection, y = prob, colour = Selection)) +
    
  geom_point(size = 2) +
  
  geom_errorbar(aes(ymin = asymp.LCL, ymax = asymp.UCL),
                width = 0.12, linewidth = 0.7) +
  
  geom_text(data = contrast_df, aes(x = 1.5, y = 1.0, label = sig),
      inherit.aes = FALSE, colour = "black", size = 5) +
    
  facet_wrap(~ dataset, nrow = 1, labeller = as_labeller(facet_labs)) +
    
  labs(title = "Isolated CG genetic effects", 
       y = "Proportion of offspring", colour = "Selection")  +
  scale_y_continuous(limits = c(0,1.05)) +
  theme_classic(base_size = 14) +
  theme(plot.title = element_text(face = "bold", hjust = 0.5, size = 16),
    strip.background = element_blank(),
    strip.text = element_text(face = "bold", size = 8),
    axis.text.x = element_text(face = "bold"),
    axis.title.y = element_text(face = "bold"),
    legend.title = element_text(face = "bold"),
    panel.spacing = unit(1, "lines"))

10.4 Assumption checks

Show assumption checks

diagnostics <- imap(cg, \(mod, nm) {
  
  model_name <- case_when(
    nm == "f_Auto" ~ "CG autosomes in females",
    nm == "m_Auto" ~ "CG autosomes in males",
    nm == "f_Cyto" ~ "CG cytoplasm in females",
    nm == "m_Cyto" ~ "CG cytoplasm in males",
    nm == "m_Y"    ~ "CG Y chromosome in males"
  )
  
  cat("\n========================\n")
  cat("Model:", model_name, "\n")
  cat("========================\n")
  
  sim_res <- simulateResiduals(mod)
  
  # residual plots
  plot(sim_res)
  
  # diagnostic tests
  disp <- testDispersion(sim_res)
  out  <- testOutliers(sim_res)
  uni  <- testUniformity(sim_res)
  
  print(disp)
  print(out)
  print(uni)
  
  list(
    sim_res = sim_res,
    dispersion = disp,
    outliers = out,
    uniformity = uni
  )
  
})

## 
## ========================
## Model: CG autosomes in females 
## ========================

## 
##  DHARMa nonparametric dispersion test via sd of residuals fitted vs.
##  simulated
## 
## data:  simulationOutput
## dispersion = 1.0591, p-value = 0.512
## alternative hypothesis: two.sided
## 
## 
##  DHARMa bootstrapped outlier test
## 
## data:  sim_res
## outliers at both margin(s) = 2, observations = 216, p-value = 0.94
## alternative hypothesis: two.sided
##  percent confidence interval:
##  0.00000000 0.02314815
## sample estimates:
## outlier frequency (expected: 0.00791666666666667 ) 
##                                        0.009259259 
## 
## 
##  Asymptotic one-sample Kolmogorov-Smirnov test
## 
## data:  simulationOutput$scaledResiduals
## D = 0.044742, p-value = 0.7802
## alternative hypothesis: two-sided
## 
## 
## ========================
## Model: CG autosomes in males 
## ========================

## 
##  DHARMa nonparametric dispersion test via sd of residuals fitted vs.
##  simulated
## 
## data:  simulationOutput
## dispersion = 0.50733, p-value < 2.2e-16
## alternative hypothesis: two.sided
## 
## 
##  DHARMa bootstrapped outlier test
## 
## data:  sim_res
## outliers at both margin(s) = 0, observations = 216, p-value = 0.36
## alternative hypothesis: two.sided
##  percent confidence interval:
##  0.00000000 0.03020833
## sample estimates:
## outlier frequency (expected: 0.0099537037037037 ) 
##                                                 0 
## 
## 
##  Asymptotic one-sample Kolmogorov-Smirnov test
## 
## data:  simulationOutput$scaledResiduals
## D = 0.055708, p-value = 0.514
## alternative hypothesis: two-sided
## 
## 
## ========================
## Model: CG cytoplasm in females 
## ========================

## 
##  DHARMa nonparametric dispersion test via sd of residuals fitted vs.
##  simulated
## 
## data:  simulationOutput
## dispersion = 0.97156, p-value = 0.624
## alternative hypothesis: two.sided
## 
## 
##  DHARMa bootstrapped outlier test
## 
## data:  sim_res
## outliers at both margin(s) = 3, observations = 209, p-value = 0.56
## alternative hypothesis: two.sided
##  percent confidence interval:
##  0.00000000 0.02643541
## sample estimates:
## outlier frequency (expected: 0.0100956937799043 ) 
##                                        0.01435407 
## 
## 
##  Asymptotic one-sample Kolmogorov-Smirnov test
## 
## data:  simulationOutput$scaledResiduals
## D = 0.035906, p-value = 0.9504
## alternative hypothesis: two-sided
## 
## 
## ========================
## Model: CG cytoplasm in males 
## ========================

## 
##  DHARMa nonparametric dispersion test via sd of residuals fitted vs.
##  simulated
## 
## data:  simulationOutput
## dispersion = 0.95513, p-value = 0.752
## alternative hypothesis: two.sided
## 
## 
##  DHARMa bootstrapped outlier test
## 
## data:  sim_res
## outliers at both margin(s) = 1, observations = 192, p-value = 0.92
## alternative hypothesis: two.sided
##  percent confidence interval:
##  0.00000000 0.03645833
## sample estimates:
## outlier frequency (expected: 0.0103645833333333 ) 
##                                       0.005208333 
## 
## 
##  Asymptotic one-sample Kolmogorov-Smirnov test
## 
## data:  simulationOutput$scaledResiduals
## D = 0.054642, p-value = 0.6152
## alternative hypothesis: two-sided
## 
## 
## ========================
## Model: CG Y chromosome in males 
## ========================

## 
##  DHARMa nonparametric dispersion test via sd of residuals fitted vs.
##  simulated
## 
## data:  simulationOutput
## dispersion = 0.91067, p-value = 0.424
## alternative hypothesis: two.sided
## 
## 
##  DHARMa bootstrapped outlier test
## 
## data:  sim_res
## outliers at both margin(s) = 0, observations = 180, p-value = 0.54
## alternative hypothesis: two.sided
##  percent confidence interval:
##  0.00000000 0.03069444
## sample estimates:
## outlier frequency (expected: 0.00844444444444445 ) 
##                                                  0 
## 
## 
##  Asymptotic one-sample Kolmogorov-Smirnov test
## 
## data:  simulationOutput$scaledResiduals
## D = 0.06064, p-value = 0.5222
## alternative hypothesis: two-sided

11 Breeding designs

11.1 Figure S1.

S1(a) ML selection breeding design. S1(b) Recombination box design.

Females denoted as DTP in the schematic represent clone-generator (CG) females carrying translocated autosomes marked with recessive eye-colour markers. Where indicated, females denoted as DTW instead represent CG females carrying translocated autosomes marked with dominant eye-colour markers.

11.2 Figure S2.

S2(a) HC males and S2(b) WT males used for CRF assays and mating success and sperm offense assays.

11.3 Figure S3.

Single artefact males carrying: S3(a) CG cytoplasm, S3(b) CG Y chromosome, and S3(c) CG autosomes

11.4 Figure S4.

Single artefact females carrying S4(a) CG cytoplasm and S4(b) CG autosomes

12 List of abbreviations

library(kableExtra)

abbr_tbl <- tibble::tribble(
  ~Factor,       ~Abbreviation, ~Meaning,
  
  "Selection",   "ML",          "Male limited",
  "Selection",   "MC",          "Matched control",
  
  "Female",      "C",           "Control",
  "Female",      "CG",          "Clone generator",
  
  "Background",  "C",           "Control",
  "Background",  "CG",          "Clone generator",
  
  "Replicate",   "1",           "Replicate 1",
  "Replicate",   "3",           "Replicate 3",
  "Replicate",   "5",           "Replicate 5"
)

abbr_tbl$Factor <- factor(abbr_tbl$Factor, levels = unique(abbr_tbl$Factor))

group_sizes <- table(abbr_tbl$Factor) 
group_sizes <- group_sizes[levels(abbr_tbl$Factor)]

abbr_tbl %>%
  select(-Factor) %>%
  kable(
    booktabs = TRUE,
    linesep = "",
    align = c("c", "l"),
    col.names = c("Abbreviation", "Description")
  ) %>%
  kable_styling(
    full_width = FALSE,
    position = "left",
    font_size = 11
  ) %>%
  pack_rows(
    index = group_sizes,
    bold = TRUE
  )

Abbreviation	Description
Selection
ML	Male limited
MC	Matched control
Female
C	Control
CG	Clone generator
Background
C	Control
CG	Clone generator
Replicate
1	Replicate 1
3	Replicate 3
5	Replicate 5

Analysis and QC - MS ID EVO-26-0131

1 CRF

1.0.1 Data import and formatting

1.1 Global model.

1.2 Contrast testing.

1.3 Dataset visualisation

1.4 Assumption checks

2 MC success

2.0.1 Data import and formatting

2.1 Global model.

2.2 Contrast testing.

2.3 Dataset visualisation

2.4 Assumption checks

3 MC latency

3.0.1 Data import and filtering

3.1 Global model

3.2 Contrast testing

3.3 Dataset visualisation

3.4 Assumption checks

4 MC duration

4.0.1 Data import and filtering

4.1 Global model

4.2 Contrast testing

4.3 Dataset visualisation

4.4 Assumption checks

5 MC fecundity induction

5.0.1 Data import and formatting

5.1 Global model

5.2 Contrast testing

5.3 Dataset visualisation

5.4 Assumption checks

6 MC Brood sex ratios

6.0.1 Data import and formatting

6.1 Global model

6.2 Contrast testing

6.3 Dataset visualisation

6.4 Assumption checks

7 SC sperm offense

7.0.1 Data import and formatting

7.1 Global model

7.2 Contrast testing

7.3 Dataset visualisation

7.4 Assumption checks

8 MH CG Mate Harm

8.0.1 Data import and formatting

8.1 Global model

8.2 Contrast testing

8.3 Dataset visualisation

8.4 Assumption checks

9 MH CG female fecundity

9.0.1 Data import and formatting

9.1 Global model

9.2 Contrast testing

9.3 Dataset visualisation

9.4 Assumption checks

10 Isolated CG genetic effects

10.0.1 Data import and formatting

10.1 Global models

10.2 Contrast testing

10.3 Dataset visualisation

10.4 Assumption checks

11 Breeding designs

11.1 Figure S1.

11.2 Figure S2.

11.3 Figure S3.

11.4 Figure S4.

12 List of abbreviations