---
title: "Memoire Laroy Ludovic"
author: "Laroy_Ludovic"
date: "2023-11-01"
output:
  html_document: # options pour sortie HTML
    code_folding: hide #  Cache le code (mais permet de l'ouvrir :-))
    collapsed: yes # Crée un document unique
    fig_caption: yes # Figures encapsulées ?
    fig_height: 5 # Hauteur par défaut des figures
    fig_width: 6 # Largeur par défaut des figures
    highlight: tango # style de mise en valeur du code
    number_sections: yes # Ajout table des matières
    theme: united  # Style du document
    toc: yes # Table des matiere
    toc_depth: 3  # Profondeur de la table des matière
    toc_float: yes # table des matière flottante
  word_document: default
  pdf_document: # options (de base) pour sorties pdf
    toc: yes
    toc_depth: '3'
editor_options:
  chunk_output_type: inline
---

# Packages and Working Directory

```{r}
#setwd("C:/Users/Ludov/Desktop/mémoire/statistiques")
```


```{r, warning=FALSE}
library(tidyverse)
library(visreg)
library(emmeans)
library(EMSaov)
library(ade4)
library(ggResidpanel)
library(car)
library(readxl)
library(lme4)
library(lmerTest)
library(knitr)
```



# Importation des données

```{r}
Data <- read_excel("data/Data_full.xlsx")
#View(Data)
```

```{r}
Data_logger_3_Dark <- read.csv2("data/Data_logger_3_Dark.csv")
Data_logger_3_Smart <- read.table("data/3_Smart.csv", header = T, sep = "\t", dec = ".")
Data_logger_3_Full <- read.table("data/3_Full.csv", header = T, sep = "\t", dec = ".")
Data_logger_4_Dark <- read.table("data/4_Dark.csv", header = T, sep = "\t", dec = ".")
Data_logger_4_Smart <- read.table("data/4_Smart.csv", header = T, sep = "\t", dec = ".")
Data_logger_4_Full <- read.table("data/4_Full.csv", header = T, sep = "\t", dec = ".")
```


## Modifications des colonnes

### Data_logger

```{r}
colnames(Data_logger_3_Dark) <- c("Sample", "Date_time", "Temperature","Humidity","Dew_point_temperature")
Data_logger_3_Dark <- Data_logger_3_Dark[2:nrow(Data_logger_3_Dark),]
Data_logger_3_Dark$Treatment <- "Dark"
Data_logger_3_Dark$Sample <- as.numeric(Data_logger_3_Dark$Sample)
Data_logger_3_Dark$Temperature <- as.numeric(Data_logger_3_Dark$Temperature)
Data_logger_3_Dark$Humidity <- as.numeric(Data_logger_3_Dark$Humidity)
Data_logger_3_Dark$Dew_point_temperature <- as.numeric(Data_logger_3_Dark$Dew_point_temperature)
Data_logger_3_Dark$Treatment <- as.factor(Data_logger_3_Dark$Treatment)
Data_logger_3_Dark$Room <- as.factor("3")
Data_logger_3_Dark$Section <- as.factor("3_Dark")
```

```{r}
colnames(Data_logger_3_Smart) <- c("Sample", "Date_time", "Temperature","Humidity","Dew_point_temperature")
Data_logger_3_Smart$Treatment <- as.factor("Smart")
Data_logger_3_Smart$Room <- as.factor("3")
Data_logger_3_Smart$Section <- as.factor("3_Smart")

```

```{r}
colnames(Data_logger_3_Full) <- c("Sample", "Date_time", "Temperature","Humidity","Dew_point_temperature")
Data_logger_3_Full$Treatment <- as.factor("Full")
Data_logger_3_Full$Room <- as.factor("3")
Data_logger_3_Full$Section <- as.factor("3_Full")
```

```{r}
colnames(Data_logger_4_Dark) <- c("Sample", "Date_time", "Temperature","Humidity","Dew_point_temperature")
Data_logger_4_Dark$Treatment <- as.factor("Dark")
Data_logger_4_Dark$Room <- as.factor("4")
Data_logger_4_Dark$Section <- as.factor("4_Dark")
```

```{r}
colnames(Data_logger_4_Smart) <- c("Sample", "Date_time", "Temperature","Humidity","Dew_point_temperature")
Data_logger_4_Smart$Treatment <- as.factor("Smart")
Data_logger_4_Smart$Room <- as.factor("4")
Data_logger_4_Smart$Section <- as.factor("4_Smart")
```

```{r}
colnames(Data_logger_4_Full) <- c("Sample", "Date_time", "Temperature","Humidity","Dew_point_temperature")
Data_logger_4_Full$Treatment <- as.factor("Full")
Data_logger_4_Full$Room <- as.factor("4")
Data_logger_4_Full$Section <- as.factor("4_Full")
```


```{r}
Data_logger <- rbind(Data_logger_3_Dark, Data_logger_3_Smart, Data_logger_3_Full, Data_logger_4_Dark, Data_logger_4_Smart, Data_logger_4_Full)
summary(Data_logger)
```


### Data

```{r, warning=FALSE}
Data$Room <- as.factor(Data$Room)
Data$Mother <- as.factor(Data$Mother)
Data$Treatment <- as.factor(Data$Treatment)
Data$Treatment_bis <- as.factor(Data$Treatment_bis)
Data$ID <- as.factor(Data$ID)
Data$Sex <- as.factor(Data$Sex)
Data$Species <- as.factor(Data$Species)
Data$Region <- as.factor(Data$Region)
Data$Site <- as.factor(Data$Site)

Data$M0 <- as.numeric(Data$M0)
Data$M1 <- as.numeric(Data$M1)
Data$M2 <- as.numeric(Data$M2)
Data$M3 <- as.numeric(Data$M3)
Data$M4 <- as.numeric(Data$M4)
Data$M5 <- as.numeric(Data$M5)
Data$M6 <- as.numeric(Data$M6)
Data$M7 <- as.numeric(Data$M7)
Data$M8 <- as.numeric(Data$M8)
Data$M9 <- as.numeric(Data$M9)
Data$M10 <- as.numeric(Data$M10)
Data$M11 <- as.numeric(Data$M11)
Data$M12 <- as.numeric(Data$M12)
Data$M13 <- as.numeric(Data$M13)
Data$M14 <- as.numeric(Data$M14)
Data$M15 <- as.numeric(Data$M15)
Data$M16 <- as.numeric(Data$M16)
Data$M17 <- as.numeric(Data$M17)
Data$M18 <- as.numeric(Data$M18)
Data$M19 <- as.numeric(Data$M19)
Data$M20 <- as.numeric(Data$M20)

Data$Time_alive <- as.numeric(Data$Time_alive)
Data$Pupa_mass <- as.numeric(Data$Pupa_mass)

Data$Did_pupated <- as.factor(Data$Did_pupated)
Data$Did_emerged <- as.factor(Data$Did_emerged)
Data$Larva_survival <- as.factor(Data$Larva_survival)
Data$Diapause <- as.factor(Data$Diapause)

Data$Treatment_FTL <- as.factor(Data$Treatment_FTL)
Data$Light_orientation <- as.factor(Data$Light_orientation)
Data$Did_move <- as.factor(Data$Did_move)
Data$Did_FTL <- as.factor(Data$Did_FTL)

Data$Treatment_bis <- factor(Data$Treatment_bis, levels = c("Dark", "Full", "Smart", "Mixt_smart"))
```

## Création des colonnes "Total_eaten" et "Eaten_per_day"

Mise en place de Total_eaten
```{r}
Vecteur_total_eaten <- numeric()
for ( Y in (1:450) ) {
  

  B <- 0
  for ( X in (seq(73,403, by = 3)) ) {
    A <- as.numeric(unname(Data[ Y , X ]))
    B <- sum( A , B )
  
  }
  
  Vecteur_total_eaten <- c(Vecteur_total_eaten, B)
  
}

Data$Total_eaten <- Vecteur_total_eaten
```


Mise en place de Eaten_per_day
```{r}
Data <- Data %>% mutate(Eaten_per_day = Total_eaten / Time_alive )
```

## Speed_to_light

Afin de résoudre prendre en compte tout les individus qui ont n'ont pas voler mais participer à l'expérience (ils ont aucun score dans le temps mis pour voler mais ne participent pas au test stat (car ils ont rien dans la colonne Flight_to_light))

```{r}
Vecteur_speed <- numeric()
for (X in (1:nrow(Data))) {
  A <- ifelse(Data$End_FTL_experiment[X] == 1800, 0, ( 4.5 / (Data$Flight_to_light[X] / 60) ))
  Vecteur_speed <- c(Vecteur_speed, A)
}

Data$Speed_to_light <- Vecteur_speed
```

## Ordre des facteurs différents
```{r}
Data_ref_smart <- Data
Data_ref_smart$Treatment <- relevel(Data$Treatment, ref = "Smart")
Data_ref_smart$Treatment_bis <- relevel(Data$Treatment_bis, ref = "Smart")

Data_ref_full <- Data
Data_ref_full$Treatment <- relevel(Data$Treatment, ref = "Full")
Data_ref_full$Treatment_bis <- relevel(Data$Treatment_bis, ref = "Full")

Data_ref_mixt_smart <- Data
Data_ref_mixt_smart$Treatment_bis <- relevel(Data$Treatment_bis, ref = "Mixt_smart")

Data_refFTL_Night <- Data
Data_refFTL_Night$Treatment_FTL <- relevel(Data$Treatment_FTL, ref = "Night")
Data_refFTL_Night$Treatment_FTL <- relevel(Data$Treatment_FTL, ref = "Night")

Data_refFTL_Mix <- Data
Data_refFTL_Mix$Treatment_FTL <- relevel(Data$Treatment_FTL, ref = "Mix")
Data_refFTL_Mix$Treatment_FTL <- relevel(Data$Treatment_FTL, ref = "Mix")
```

## Séparation des jeux de données en fonction par espèce
```{r}
Data_OP <- Data %>% filter(Species == "OP")
Data_OP_ref_smart <- Data_ref_smart %>% filter(Species == "OP")
Data_OP_ref_full  <- Data_ref_full  %>% filter(Species == "OP")

Data_AE <- Data %>% filter(Species == "AE")
Data_AE_ref_smart <- Data_ref_smart %>% filter(Species == "AE")
Data_AE_ref_full  <- Data_ref_full  %>% filter(Species == "AE")
Data_AE_ref_mixt_smart  <- Data_ref_mixt_smart  %>% filter(Species == "AE")
```


# Data logger

```{r}
summary(Data_logger)
```


## Température par pièce
```{r}
Anova_Data_logger_temp_room <- aov(Temperature ~ Room, data = Data_logger)
summary(Anova_Data_logger_temp_room)
```

```{r}
kable(TukeyHSD(Anova_Data_logger_temp_room)$Room,caption = "Test post-hoc de Tukey pour la température en fonction des différentes pièces")
write.csv2( TukeyHSD(Anova_Data_logger_temp_room)$Room , file = "output/Data_logger_temp_room.csv" )
```

## Température par traitement

```{r}
Anova_Data_logger_temp_treatment <- aov(Temperature ~ Treatment, data = Data_logger)
summary(Anova_Data_logger_temp_treatment)
```

```{r}
kable(TukeyHSD(Anova_Data_logger_temp_treatment)$Treatment,caption = "Test post-hoc de Tukey pour la température en fonction des différents traitements")
write.csv2( TukeyHSD(Anova_Data_logger_temp_treatment)$Treatment , file = "output/Data_logger_temp_treatment.csv" )
```


## Température par section
```{r}
Anova_Data_logger_temp_section <- aov(Temperature ~ Section, data = Data_logger)
summary(Anova_Data_logger_temp_section)
```

```{r}
kable(TukeyHSD(Anova_Data_logger_temp_section)$Section,caption = "Test post-hoc de Tukey pour la température en fonction des différentes sections")
write.csv2( TukeyHSD(Anova_Data_logger_temp_section)$Section , file = "output/Data_logger_temp_section.csv" )
```

## Humidité par pièce
```{r}
Anova_Data_logger_hum_room <- aov(Humidity ~ Room, data = Data_logger)
summary(Anova_Data_logger_hum_room)
```

```{r}
kable(TukeyHSD(Anova_Data_logger_hum_room)$Room,caption = "Test post-hoc de Tukey pour l'humidité en fonction des différentes pièces")
write.csv2( TukeyHSD(Anova_Data_logger_hum_room)$Room , file = "output/Data_logger_hum_room.csv" )
```

## Humidité par traitement

```{r}
Anova_Data_logger_hum_treatment <- aov(Humidity ~ Treatment, data = Data_logger)
summary(Anova_Data_logger_hum_treatment)
```

```{r}
kable(TukeyHSD(Anova_Data_logger_hum_treatment)$Treatment,caption = "Test post-hoc de Tukey pour l'humidité en fonction des différents traitements")
write.csv2( TukeyHSD(Anova_Data_logger_hum_treatment)$Treatment , file = "output/Data_logger_hum_treatment.csv" )
```


## Humidité par section
```{r}
Anova_Data_logger_hum_section <- aov(Humidity ~ Section, data = Data_logger)
summary(Anova_Data_logger_hum_section)
```

```{r}
kable(TukeyHSD(Anova_Data_logger_hum_section)$Section,caption = "Test post-hoc de Tukey pour l'humidité en fonction des différentes sections")
write.csv2( TukeyHSD(Anova_Data_logger_hum_section)$Section , file = "output/Data_logger_hum_section.csv" )
```




# Analyses Développement

## Total_molt

Vu que c'est une variable de comptage, nous comptons faire un glm de type poisson

### Test

```{r}
GLM_Total_molt <- glmer(Total_molt ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex , family = poisson, data = Data)
summary(GLM_Total_molt)
```
Pas d'effet significatif autre que l'espèce.

Vérification des hypothèses
```{r}
resid_panel(GLM_Total_molt, plots = c("resid"), smoother = TRUE)
```

Le modèle pourrait être siplifié par AIC (surtout pour les interactions)

### Sipmplification du modèle

Simplification par AIC/BIC
```{r}
drop1(GLM_Total_molt)
```
```{r}
GLM_Total_molt2 <- update(GLM_Total_molt, . ~ . - Treatment_bis:Sex)
drop1(GLM_Total_molt2)
```
```{r,cache=TRUE}
GLM_Total_molt3 <- update(GLM_Total_molt2, . ~ . - Treatment_bis:Species)
drop1(GLM_Total_molt3)
```

```{r,cache=TRUE}
GLM_Total_molt4 <- update(GLM_Total_molt3, . ~ . - Sex:Species)
# drop1(GLM_Total_molt5) drop1 a cessé de fonctionné donc je compare le modèle avec et  sans interactions
GLM_Total_molt5 <- update(GLM_Total_molt4, . ~ . - Species:Region - Room:Treatment_bis)
AIC(GLM_Total_molt4)
AIC(GLM_Total_molt5)
```

Modèle final, aucune interactions n'était significative
```{r}
GLM_Total_molt <- GLM_Total_molt5
summary(GLM_Total_molt5)
```
Seulement les espèces sont significatives

### Treatment VS Treatment_bis

```{r}
AIC(GLM_Total_molt)
AIC(update(GLM_Total_molt, . ~ . - Treatment_bis + Treatment))
```
```{r}
summary(GLM_Total_molt)$coeff
summary(update(GLM_Total_molt, . ~ . - Treatment_bis + Treatment))$coeff
```
pas de grosse différence de significativité
L'AIC préfère le modèle avec "Treatment"


```{r}
summary(GLM_Total_molt)
summary(update(GLM_Total_molt, . ~ . - Treatment_bis + Treatment))
```




```{r}
Info_Total_molt <- "
Effet significatif de l'espèce, OP possède moins de mues.
"
Info_Total_molt
```


### Visualisation

Résumé des données
```{r, message=FALSE}
Summary_Total_molt <- Data %>% 
  group_by(Treatment, Species, Region) %>% 
  summarise(Mean=mean(Total_molt),
            Var=var(Total_molt),
            N=n()) %>% 
  mutate(CI_S=Mean+qt(0.975, N-1)*sqrt(Var/N),
         CI_I=Mean-qt(0.975, N-1)*sqrt(Var/N))
```


Graphique avec ce résumé
```{r}
ggplot(Summary_Total_molt, aes(x=Treatment, y=Mean, color = Species))+
  geom_errorbar(aes(ymin=CI_I, ymax=CI_S),width=0.2,size = 1.2, alpha = 0.5) +
  geom_point(alpha = 0.8) +
  facet_grid(~Region) +
  labs(title = "Nombre de mues", x = "Traitement", y = "Nombre de mues", color = "Espèce") +
  theme_bw()
```

```{r}
ggplot(Data, aes(x=Total_molt)) +
  geom_histogram(aes(y = ..count.. , fill = Species), position = "dodge", size = 4) +
  scale_x_continuous(breaks = seq(0,6)) +
  labs(title = "Nombres de mues par espèce", x = "Nombre de mues", y = "Nombre d'individu", fill = "Espèce") +
  theme_bw()
```




## Diapause

### Test

```{r, warning = F}
GLM_Diapause <- glmer(Diapause ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data, family = binomial)
summary(GLM_Diapause)
```

### Simplification du modèle

```{r, warning = F}
drop1(GLM_Diapause)
```
```{r,warning=FALSE, message=FALSE}
GLM_Diapause1 <- update(GLM_Diapause, . ~ . - Treatment_bis:Region)
drop1(GLM_Diapause1)
```
Plus aucune interaction est nécessaire à faire retirer

```{r}
GLM_Diapause <- GLM_Diapause1
```

### Treatment VS Treatment_bis

```{r, warning = F}
AIC(glmer(Diapause ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data, family = binomial))
AIC(glmer(Diapause ~ Treatment + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data, family = binomial))
```

```{r, warning = F}
summary(glmer(Diapause ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data, family = binomial))$coeff
summary(glmer(Diapause ~ Treatment + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data, family = binomial))$coeff
```

Aucune grosse différence, on peut continuer avec Treatment_bis (qui est meilleur en termes d'AIC)


Analyse des p-val

```{r, warning = F}
summary(glmer(Diapause ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data, family = binomial))
summary(glmer(Diapause ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_ref_smart, family = binomial))
summary(glmer(Diapause ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_ref_full, family = binomial))
```


```{r}
Info_Diapause <-"
Les individus du traitement Full ont quasiements significativement (p-val = 0.0728) une plus grande probabilité d'être en diapause que les individus du traitement Dark.
Les mâles du traitement Full ont significativements une plus petite probabilité de rentrer en diapause par rapport aux femelles du même traitement.
"
```


### Visualisation

```{r}
ggplot(Data) +
  aes(x = Diapause, fill = Treatment) +
  geom_bar(position = "dodge") +
  scale_fill_hue(direction = 1) +
  facet_grid(~Species) +
  labs(title = "Nombre de diapause chez AE et OP", y = "Nombre", x = "Diapause", fill = "Traitement") +
  theme_minimal()
```





## TT_pupation


### Test

```{r}
LM_TT_pupation <- lmer(TT_pupation ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex + Treatment_bis:Region, data = Data)
summary(LM_TT_pupation)
```

Normalité des résidus
```{r}
qqnorm(resid(LM_TT_pupation))
qqline(resid(LM_TT_pupation))
```
Normalité non vérifiée, on modifie les données

```{r}
LM_TT_pupation <- lmer(log(TT_pupation) ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex + Treatment_bis:Region, data = Data)
```
Normalité des résidus
```{r}
qqnorm(resid(LM_TT_pupation))
qqline(resid(LM_TT_pupation))
```
C'est bien mieux

Homogénité des variances
```{r}
plot(LM_TT_pupation)
```

Distance de cook
```{r}
plot(cooks.distance(LM_TT_pupation))
```


### Simplification du modèle

```{r}
drop1(LM_TT_pupation)
```
```{r}
LM_TT_pupation1 <- update(LM_TT_pupation, . ~ . - Treatment_bis:Sex)
drop1(LM_TT_pupation1)
```
```{r}
LM_TT_pupation2 <- update(LM_TT_pupation1, . ~ . - Species:Region)
drop1(LM_TT_pupation2)
```
```{r}
LM_TT_pupation3 <- update(LM_TT_pupation2, . ~ . - Sex:Species)
drop1(LM_TT_pupation3)
```
```{r}
LM_TT_pupation4 <- update(LM_TT_pupation3, . ~ . - Sex)
drop1(LM_TT_pupation4)
```

```{r}
AIC(LM_TT_pupation)
AIC(LM_TT_pupation2)
AIC(LM_TT_pupation3)
AIC(LM_TT_pupation4)
```
Le meilleur modèle est le dernier (et en plus c'est le plus simple)
```{r}
LM_TT_pupation <- lmer(log(TT_pupation) ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex , data = Data)
```




### Treatment VS Treatment_bis
```{r}
AIC(lmer(log(TT_pupation) ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex , data = Data))
AIC(lmer(log(TT_pupation) ~ Treatment + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment:Species + Treatment:Sex , data = Data))
```

```{r}
summary(lmer(log(TT_pupation) ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Region + Treatment_bis:Species , data = Data))$coeff
summary(lmer(log(TT_pupation) ~ Treatment + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment:Region + Treatment:Species, data = Data))$coeff
```
Le modèle treatment_bis a des p-val concernants les traitements plus significatifs et est mieux selon l'AIC.

Analyse des P-val
```{r}
summary(lmer(log(TT_pupation) ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Region + Treatment_bis:Species , data = Data))
summary(lmer(log(TT_pupation) ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Region + Treatment_bis:Species , data = Data_ref_full))
summary(lmer(log(TT_pupation) ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Region + Treatment_bis:Species , data = Data_ref_smart))
```

```{r}
Infos_TT_pupation <- "
OP est plus rapide.
Smart est significativement plus rapide que Dark chez AE mais Smart est significativement plus lent que Dark et Full chez OP.
Full est quasiment significativement (0.082) plus lent que Dark mais uniquement dans le Luxembourg.
Full est quasiment significativement (0.086) plus lent que Smart.
Full est quasiment significativement (0.097) plus lent dans le Luxembourg.
Mixt_smart est plus lent que n'importe quel autre treatment mais uniquement dans le Luxembourg
"
Infos_TT_pupation
```



### Visualisation

```{r, message=FALSE}
Summary_TT_pupation <- Data %>% 
  filter(Did_pupated == "Yes") %>% 
  group_by(Treatment_bis, Species, Region) %>% 
  summarise(Mean=mean(log(TT_pupation)),
            Var=var(log(TT_pupation)),
            N=n()) %>% 
  mutate(CI_S=Mean+qt(0.975, N-1)*sqrt(Var/N),
         CI_I=Mean-qt(0.975, N-1)*sqrt(Var/N))
```


```{r}
ggplot(Summary_TT_pupation, aes(x=Treatment_bis, y=Mean, color = Species))+
  geom_errorbar(aes(ymin=CI_I, ymax=CI_S),width=0.2,size = 1.2, alpha = 0.5) +
  geom_point(alpha = 0.8) +
  facet_grid(~Region) +
  labs(title = "Temps pour entrer en chrysalide", x = "Traitement", y = "log(Nombre de jours)", color = "Espèce") +
  theme_bw()
```



```{r, message=FALSE}
Summary_TT_pupation2 <- Data %>%
  filter(Did_pupated == "Yes") %>% 
  group_by(Treatment_bis, Species) %>% 
  summarise(Mean=mean(log(TT_pupation)),
            Var=var(log(TT_pupation)),
            N=n()) %>% 
  mutate(CI_S=Mean+qt(0.975, N-1)*sqrt(Var/N),
         CI_I=Mean-qt(0.975, N-1)*sqrt(Var/N))
```


```{r}
ggplot(Summary_TT_pupation2, aes(x=Treatment_bis, y=Mean, color = Species))+
  geom_errorbar(aes(ymin=CI_I, ymax=CI_S),width=0.2,size = 1.2, alpha = 0.5) +
  geom_point(alpha = 0.8) +
  labs(title = "Time to Pupation", x = "Treatment", y = "log(Days)") +
  theme_bw()
```


## Time_in_pupa

### Test

```{r}
LM_Time_in_pupa <- lmer(Time_in_pupa ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex + Treatment_bis:Region, data = Data)
summary(LM_TT_pupation)
```

Vérification de la normalité
```{r}
qqnorm(resid(LM_Time_in_pupa))
qqline(resid(LM_Time_in_pupa))
```
Homogénité des variances
```{r}
plot(LM_Time_in_pupa)
```

Distance de cook
```{r}
plot(cooks.distance(LM_Time_in_pupa))
```

### Simplification du modèle

```{r}
drop1(LM_Time_in_pupa)
```
```{r}
LM_Time_in_pupa1 <- update(LM_Time_in_pupa, . ~ . - Treatment_bis:Species)
drop1(LM_Time_in_pupa1)
```
```{r}
LM_Time_in_pupa2 <- update(LM_Time_in_pupa1, . ~ . - Sex:Species)
drop1(LM_Time_in_pupa2)
```
```{r}
LM_Time_in_pupa3 <- update(LM_Time_in_pupa2, . ~ . - Treatment_bis:Sex)
drop1(LM_Time_in_pupa3)
```
```{r}
LM_Time_in_pupa4 <- update(LM_Time_in_pupa3, . ~ . - Species:Region)
drop1(LM_Time_in_pupa4)
```
Regardons si les simplifications étaient utiles
```{r}
AIC(LM_Time_in_pupa4)
AIC(LM_Time_in_pupa)
```
Les simplifications étaient utiles
```{r}
LM_Time_in_pupa <- LM_Time_in_pupa4
```


### Treatment VS Treatment_bis

```{r}
AIC(lmer(Time_in_pupa ~ Treatment_bis + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region , data = Data))
AIC(lmer(Time_in_pupa ~ Treatment + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment:Region , data = Data))
```
```{r}
summary(lmer(Time_in_pupa ~ Treatment_bis + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region , data = Data))$coeff
summary(lmer(Time_in_pupa ~ Treatment + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment:Region , data = Data))$coeff
```

Selon les summary le choix de treatment ou treatment_bis est discutable mais on peut alors se pencher sur l'AIC qui nous encourage à prendre treatment_bis.

Analysons les P-val
```{r}
summary(LM_Time_in_pupa)
summary(lmer(Time_in_pupa ~ Treatment_bis + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region , data = Data_ref_full))
summary(lmer(Time_in_pupa ~ Treatment_bis + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region , data = Data_ref_smart))
```

```{r}
Info_time_in_pupa <- "
Le traitement Full met significativement plus de temps que le traitement Dark.
Le traitement Smart met significativement moins de temps que le traitement Full.
Le traitement Smart a une p-val de 0.999 quand on le compare à dark. Ces 2 traitements sont donc très simillaires.
Les males sont significatiment plus lent
L'espèce OP est significativement plus rapide
Les chenilles restent plus longtemps en chrysalide dans le traitement Mixt_smart quand elles viennent du luxembourg que quand elles viennent du brabant.
"
Info_time_in_pupa
```


### Visualisation


```{r, message=FALSE}
Summary_Time_in_pupa <- Data %>% 
  filter(Did_pupated == "Yes", Did_emerged == "Yes", Sex != "NA") %>% 
  group_by(Treatment_bis, Species, Region) %>% 
  summarise(Mean=mean(Time_in_pupa),
            Var=var(Time_in_pupa),
            N=n()) %>% 
  mutate(CI_S=Mean+qt(0.975, N-1)*sqrt(Var/N),
         CI_I=Mean-qt(0.975, N-1)*sqrt(Var/N))
```


```{r}
ggplot(Summary_Time_in_pupa, aes(x=Treatment_bis, y=Mean, color = Species))+
  geom_errorbar(aes(ymin=CI_I, ymax=CI_S),width=0.2,size = 1.2, alpha = 0.5) +
  geom_point(alpha = 0.8) +
  facet_wrap(~Region) +
  labs(title = "Temps en chrysalide", x = "Traitement", y = "Nombre de jours", color = "Espèce") +
  theme_bw()
```
Tout se voit super bien sur ce graph sauf le Sex.

```{r, message=FALSE}
Summary_Time_in_pupa2 <- Data %>% 
  filter(Did_pupated == "Yes", Did_emerged == "Yes", Sex != "NA") %>% 
  group_by(Treatment_bis, Species, Sex) %>% 
  summarise(Mean=mean(Time_in_pupa),
            Var=var(Time_in_pupa),
            N=n()) %>% 
  mutate(CI_S=Mean+qt(0.975, N-1)*sqrt(Var/N),
         CI_I=Mean-qt(0.975, N-1)*sqrt(Var/N))
```

```{r}
ggplot(Summary_Time_in_pupa2, aes(x=Treatment_bis, y=Mean, color = Sex))+
  scale_color_manual(name = "Sex", values = c("Female" = "red", "Male" = "royalblue")) +
  geom_errorbar(aes(ymin=CI_I, ymax=CI_S),width=0.2,size = 1.2, alpha = 0.5) +
  geom_point(alpha = 0.8) +
  facet_wrap(~Species) +
  labs(title = "Temps en chrysalide", x = "Traitement", y = "Nombre de jours", color = "Espèce") +
  theme_bw()
```

## TT_emerge

### Test

```{r}
LM_TT_emerge <- lmer(TT_emerge ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex + Treatment_bis:Region, data = Data)
summary(LM_TT_pupation)
```


Vérification de la normalité
```{r}
qqnorm(resid(LM_TT_emerge))
qqline(resid(LM_TT_emerge))
```

il faudrait mettre d'autres facteur
```{r}
LM_TT_emerge <- lmer(log(TT_emerge) ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex + Treatment_bis:Region, data = Data)
summary(LM_TT_pupation)
```

```{r}
qqnorm(resid(LM_TT_emerge))
qqline(resid(LM_TT_emerge))
```
La normalité est bien meilleure

Homogénité des variances
```{r}
plot(LM_TT_emerge)
```

Distance de cook
```{r}
plot(cooks.distance(LM_TT_emerge))
```

### Simplification du modèle


```{r}
drop1(LM_TT_emerge)
```
```{r}
LM_TT_emerge1 <- update(LM_TT_emerge, . ~ . - Treatment_bis:Sex)
drop1(LM_TT_emerge1)
```
```{r}
LM_TT_emerge2 <- update(LM_TT_emerge1, . ~ . - Sex:Species)
drop1(LM_TT_emerge2)
```
```{r}
LM_TT_emerge3 <- update(LM_TT_emerge2, . ~ . - Sex)
drop1(LM_TT_emerge3)
```

Regardons si les simplifications étaient utiles
```{r}
AIC(LM_TT_emerge)
AIC(LM_TT_emerge3)
```
Les simplifications étaient utiles
```{r}
LM_TT_emerge <- LM_TT_emerge3
```



### Treatment VS Treatment_bis
```{r}
AIC(lmer(log(TT_emerge) ~ Treatment_bis + Species + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Species:Region + Treatment_bis:Species + Treatment_bis:Region, data = Data))
AIC(lmer(log(TT_emerge) ~ Treatment + Species + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Species:Region + Treatment:Species + Treatment:Region, data = Data))
```
```{r}
summary(lmer(log(TT_emerge) ~ Treatment_bis + Species + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Species:Region + Treatment_bis:Species + Treatment_bis:Region, data = Data))$coeff
summary(lmer(log(TT_emerge) ~ Treatment + Species + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Species:Region + Treatment:Species + Treatment:Region, data = Data))$coeff
```
Treatment_bis est meilleur, il explique mieux les résultats obtenus

Analysons les P-val

```{r}
summary(LM_TT_emerge)
summary(lmer(log(TT_emerge) ~ Treatment_bis + Species + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Species:Region + Treatment_bis:Species + Treatment_bis:Region, data = Data_ref_full))
summary(lmer(log(TT_emerge) ~ Treatment_bis + Species + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Species:Region + Treatment_bis:Species + Treatment_bis:Region, data = Data_ref_smart))
```


```{r}
Infos_TT_emerge <- "
L'impacte du traitement Smart chez OP est quasiment significativemt (p = 0.051) différent (plus lent que Dark) que l'impacte du traitement smart chez AE.
L'impacte du traitement Smart chez OP est significativemt différent (plus lent que Full) que l'impacte du traitement smart chez AE.
Chez AE, Le traitement smart est également significativement plus rapide que le traitement full.
L'impacte du traitement Mixt_Smart est différent (plus lent) en fonction de la région
L'espèce OP met significativement moins de temps.
La région luxembourg met significativement plus de temps pour l'espèce AE mais met significativement moins de temps pour l'espèce OP.
"
Infos_TT_emerge
```


### Visualisation

```{r, message=FALSE}
Summary_TT_emerge <- Data %>% 
  filter(Did_emerged == "Yes") %>% 
  group_by(Treatment_bis, Species, Region) %>% 
  summarise(Mean=mean(log(TT_emerge)),
            Var=var(log(TT_emerge)),
            N=n()) %>% 
  mutate(CI_S=Mean+qt(0.975, N-1)*sqrt(Var/N),
         CI_I=Mean-qt(0.975, N-1)*sqrt(Var/N))
```


```{r}
ggplot(Summary_TT_emerge, aes(x=Treatment_bis, y=Mean, color = Species))+
  geom_errorbar(aes(ymin=CI_I, ymax=CI_S),width=0.2,size = 1.2, alpha = 0.5) +
  geom_point(alpha = 0.8) +
  facet_grid(~Region) +
  labs(title = "Time to Emerge", x = "Treatment", y = "log(Days)") +
  theme_bw()
```

```{r}
ggplot(Summary_TT_emerge, aes(x=Region, y=Mean, color = Species))+
  geom_errorbar(aes(ymin=CI_I, ymax=CI_S),width=0.2,size = 1.2, alpha = 0.5) +
  geom_point(alpha = 0.8) +
  facet_grid(~Treatment_bis) +
  labs(title = "Temps pour devenir adulte", x = "Région", y = "log(Nombre de jours)", color = "Espèce") +
  theme_bw()
```

```{r, message=FALSE}
Summary_TT_emerge2 <- Data %>% 
  filter(Did_emerged == "Yes") %>% 
  group_by(Treatment_bis, Species) %>% 
  summarise(Mean=mean(log(TT_emerge)),
            Var=var(log(TT_emerge)),
            N=n()) %>% 
  mutate(CI_S=Mean+qt(0.975, N-1)*sqrt(Var/N),
         CI_I=Mean-qt(0.975, N-1)*sqrt(Var/N))
```


```{r}
ggplot(Summary_TT_emerge2, aes(x=Species, y=Mean, color = Species))+
  geom_errorbar(aes(ymin=CI_I, ymax=CI_S),width=0.2,size = 1.2, alpha = 0.5) +
  geom_point(alpha = 0.8) +
  facet_grid(~Treatment_bis) +
  labs(title = "Time to Emerge", x = "Treatment", y = "log(Days)") +
  theme_bw()
```


## Pupa_mass

### Test

```{r}
LM_Pupa_mass <- lmer(Pupa_mass ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex + Treatment_bis:Region, data = Data)
summary(LM_Pupa_mass)
```

Vérification des hypothèses

```{r}
qqnorm(resid(LM_Pupa_mass))
qqline(resid(LM_Pupa_mass))
```
Une modification des données est nécessaire

```{r}
LM_Pupa_mass <- lmer(log(Pupa_mass) ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex + Treatment_bis:Region, data = Data)
summary(LM_Pupa_mass)
```

```{r}
qqnorm(resid(LM_Pupa_mass))
qqline(resid(LM_Pupa_mass))
```
Je pense que la normalité est ok.

Homogénité des variances
```{r}
plot(LM_Pupa_mass)
```

Distance de cook
```{r}
plot(cooks.distance(LM_Pupa_mass))
```

### Simplification du modèle

```{r}
drop1(LM_Pupa_mass)
```
```{r}
LM_Pupa_mass1 <- update(LM_Pupa_mass, . ~ . - Treatment_bis:Region)
drop1(LM_Pupa_mass1)
```
```{r}
LM_Pupa_mass2 <- update(LM_Pupa_mass1, . ~ . - Treatment_bis:Species)
drop1(LM_Pupa_mass2)
```
```{r}
LM_Pupa_mass3 <- update(LM_Pupa_mass2, . ~ . - Treatment_bis:Sex)
drop1(LM_Pupa_mass3)
```
```{r}
LM_Pupa_mass4 <- update(LM_Pupa_mass3, . ~ . - Sex:Species)
drop1(LM_Pupa_mass4)
```
```{r}
LM_Pupa_mass5 <- update(LM_Pupa_mass4, . ~ . - Sex)
drop1(LM_Pupa_mass5)
```
```{r}
LM_Pupa_mass6 <- update(LM_Pupa_mass5, . ~ . - Species:Region)
drop1(LM_Pupa_mass6)
```
Il n'y a plus aucune interaction, nous arrêtons la simplification du modèle

Regardons si les simplifications étaient utiles
```{r}
AIC(LM_Pupa_mass6)
AIC(LM_Pupa_mass)
```
Elle l'a été

```{r}
LM_Pupa_mass <- LM_Pupa_mass6
```

### Treatment VS Treatment_bis

```{r}
AIC(lmer(log(Pupa_mass) ~ Treatment_bis + Species + Region + (1 | Mother) +  (1 | Site) + (1 | Room), data = Data))
AIC(lmer(log(Pupa_mass) ~ Treatment + Species + Region + (1 | Mother) +  (1 | Site) + (1 | Room), data = Data))
```
```{r}
summary(lmer(log(Pupa_mass) ~ Treatment_bis + Species + Region + (1 | Mother) +  (1 | Site) + (1 | Room), data = Data))$coeff
summary(lmer(log(Pupa_mass) ~ Treatment + Species + Region + (1 | Mother) +  (1 | Site) + (1 | Room), data = Data))$coeff
```
Les AIC sont assez proche mais pour une fois je serais plus d'avis de sélectionner la colonne Treatment et non treatment_bis au vu des p-val. Je pars sur Treatment.

Analyse des p-val

```{r}
summary(lmer(log(Pupa_mass) ~ Treatment + Species + Region + (1 | Mother) +  (1 | Site) + (1 | Room), data = Data))
summary(lmer(log(Pupa_mass) ~ Treatment + Species + Region + (1 | Mother) +  (1 | Site) + (1 | Room), data = Data_ref_full))
summary(lmer(log(Pupa_mass) ~ Treatment + Species + Region + (1 | Mother) +  (1 | Site) + (1 | Room), data = Data_ref_smart))
```


```{r}
Info_Pupa_mass <- "
Le traitement Full (P = 0.0658) est presque significativement plus lourd que le le traitement Dark.
Le traitement Smart est significativement plus lourd que le traitement Dark.
L'espèce OP est significativement plus légère.
"
Info_Pupa_mass
```


### Visualisation

```{r, message=FALSE}
Summary_Pupa_mass <- Data %>% 
  filter(Pupa_mass != "NA") %>% 
  group_by(Treatment, Species, Region) %>% 
  summarise(Mean=mean(log(Pupa_mass)),
            Var=var(log(Pupa_mass)),
            N=n()) %>% 
  mutate(CI_S=Mean+qt(0.975, N-1)*sqrt(Var/N),
         CI_I=Mean-qt(0.975, N-1)*sqrt(Var/N))
```


```{r}
ggplot(Summary_Pupa_mass, aes(x=Treatment, y=Mean, color = Species))+
  geom_errorbar(aes(ymin=CI_I, ymax=CI_S),width=0.2,size = 1.2, alpha = 0.5) +
  geom_point(alpha = 0.8) +
  facet_grid(~Region) +
  labs(title = "Poids des chrysalides", x = "Traitement", y = "log(Poids des chrysalides (g) )", color = "Espèce") +
  theme_bw()
```

## Adult_mass

### Test

```{r}
LM_Adult_mass <- lmer(Adult_mass ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex + Treatment_bis:Region, data = Data)
summary(LM_Adult_mass)
```

Vérification des hypothèses

```{r}
qqnorm(resid(LM_Adult_mass))
qqline(resid(LM_Adult_mass))
```
Une modification des données est nécessaire.

```{r}
LM_Adult_mass <- lmer(log(Adult_mass) ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex + Treatment_bis:Region, data = Data)
summary(LM_Adult_mass)
```

```{r}
qqnorm(resid(LM_Adult_mass))
qqline(resid(LM_Adult_mass))
```

Homogénité des variances
```{r}
plot(LM_Adult_mass)
```

Distance de cook
```{r}
plot(cooks.distance(LM_Adult_mass))
```


### Simplification du modèle

```{r}
drop1(LM_Adult_mass)
```
```{r}
LM_Adult_mass1 <- update(LM_Adult_mass, . ~ . - Treatment_bis:Species)
drop1(LM_Adult_mass1)
```
```{r}
LM_Adult_mass2 <- update(LM_Adult_mass1, . ~ . - Treatment_bis:Region)
drop1(LM_Adult_mass2)
```
```{r}
LM_Adult_mass3 <- update(LM_Adult_mass2, . ~ . - Treatment_bis:Sex)
drop1(LM_Adult_mass3)
```
```{r}
LM_Adult_mass4 <- update(LM_Adult_mass3, . ~ . - Sex:Species)
drop1(LM_Adult_mass4)
```
```{r}
LM_Adult_mass5 <- update(LM_Adult_mass4, . ~ . - Species:Region)
drop1(LM_Adult_mass5)
```
Nous n'avons plus d'interaction, nous arrêtons la simplification du modèle.

Est ce que la simplification a été utile ?
```{r}
AIC(LM_Adult_mass5)
AIC(LM_Adult_mass)
```
Oui, clairement

```{r}
LM_Adult_mass <- LM_Adult_mass5
```


### Treatment VS Treatment_bis

```{r}
AIC(lmer(log(Adult_mass) ~ Treatment_bis + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room), data = Data))
AIC(lmer(log(Adult_mass) ~ Treatment + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room), data = Data))
```



```{r}
summary(lmer(log(Adult_mass) ~ Treatment_bis + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room), data = Data))$coeff
summary(lmer(log(Adult_mass) ~ Treatment + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room), data = Data))$coeff
summary(lmer(log(Adult_mass) ~ Treatment + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room), data = Data_ref_smart))$coeff
```

Les AIC sont plus ou moins simillaires mais vu qu'il y a une différence significative entre Mixt_smart et Smart je pense qu'il est plus correcte scientifiquement parlant de prendre Treatment_bis

Analyse des P-val
```{r}
summary(lmer(log(Adult_mass) ~ Treatment_bis + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room), data = Data))
summary(lmer(log(Adult_mass) ~ Treatment_bis + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room), data = Data_ref_full))
summary(lmer(log(Adult_mass) ~ Treatment_bis + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room), data = Data_ref_smart))
```

```{r}
Info_Adult_mass <- "
classsement croissant de poids : Dark < Smart < Full et Mixt_smart
Les males sont moins lourds
OP est moins lourds
"
Info_Adult_mass
```


### Visualisation

```{r, message=FALSE}
Summary_Adult_mass <- Data %>% 
  filter(Adult_mass != "NA", Sex != "NA") %>% 
  group_by(Treatment_bis, Species, Sex) %>% 
  summarise(Mean=mean(log(Adult_mass)),
            Var=var(log(Adult_mass)),
            N=n()) %>% 
  mutate(CI_S=Mean+qt(0.975, N-1)*sqrt(Var/N),
         CI_I=Mean-qt(0.975, N-1)*sqrt(Var/N))
```


```{r}
ggplot(Summary_Adult_mass, aes(x=Treatment_bis, y=Mean, color = Species))+
  geom_errorbar(aes(ymin=CI_I, ymax=CI_S),width=0.2,size = 1.2, alpha = 0.5) +
  geom_point(alpha = 0.8) +
  facet_grid(~Sex) +
  labs(title = "Poids des adultes", x = "Traitement", y = "log(Poids des adultes (g))", color = "Espèce") +
  theme_bw()
```

## M0

### Test

```{r}
LM_M0 <- lmer(M0 ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex + Treatment_bis:Region, data = Data)
summary(LM_M0)
```

Vérification des hypothèses

```{r}
qqnorm(resid(LM_M0))
qqline(resid(LM_M0))
```
Je pense que ca va


Homogénité des variances
```{r}
plot(LM_M0)
```
Il faudrait quand même un sqrt ou quelque chose dans le style

```{r}
LM_M0 <- lmer(sqrt(M0) ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex + Treatment_bis:Region, data = Data)
summary(LM_M0)
```

```{r}
qqnorm(resid(LM_M0))
qqline(resid(LM_M0))
```
C'est NICKEL

Homogénité des variances
```{r}
plot(LM_M0)
```
c'est correcte

Distance de cook
```{r}
plot(cooks.distance(LM_M0))
```


### Simplification du modèle

```{r}
drop1(LM_M0)
```
```{r}
LM_M0_1 <- update(LM_M0, . ~ . - Treatment_bis:Sex)
drop1(LM_M0_1)
```
```{r}
LM_M0_2 <- update(LM_M0_1, . ~ . - Treatment_bis:Species)
drop1(LM_M0_2)
```
```{r}
LM_M0_3 <- update(LM_M0_2, . ~ . - Treatment_bis:Region)
drop1(LM_M0_3)
```
```{r}
LM_M0_4 <- update(LM_M0_3, . ~ . - Sex:Species)
drop1(LM_M0_4)
```
```{r}
LM_M0_5 <- update(LM_M0_4, . ~ . - Species:Region)
drop1(LM_M0_5)
```

```{r}
summary(LM_M0_5)
```

Nous n'avons plus d'interactions nous pouvons nous arrêter là. Vu qu'il y a aucune différence significative peut-importe le modèle à part pour l'espèce (ce qui est logique), nous pouvons nous arrêter là. Notons cependant que l'espèce OP est significativement plus lourde au départ de l'expérience et que nous avions presque un effet significatif du traitement Smart.


## Total_eaten

### Test

```{r}
LM_Total_eaten <- lmer(Total_eaten ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex + Treatment_bis:Region, data = Data)
summary(LM_Total_eaten)
```

Vérification des hypothèses
```{r}
qqnorm(resid(LM_Total_eaten))
qqline(resid(LM_Total_eaten))
```
Il faut modifier les variables

```{r}
LM_Total_eaten <- lmer(log(Total_eaten) ~ Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex + Treatment_bis:Region, data = Data)
summary(LM_Total_eaten)
```

```{r}
qqnorm(resid(LM_Total_eaten))
qqline(resid(LM_Total_eaten))
```
C'est très bon.

Homogénité des variances
```{r}
plot(LM_Total_eaten)
```
c'est correcte

Distance de cook
```{r}
plot(cooks.distance(LM_Total_eaten))
```

### Simplification du modèle


```{r}
drop1(LM_Total_eaten)
```
```{r}
LM_Total_eaten1 <- update(LM_Total_eaten, . ~ . - Species:Region)
drop1(LM_Total_eaten1)
```
```{r}
LM_Total_eaten2 <- update(LM_Total_eaten1, . ~ . - Treatment_bis:Sex)
drop1(LM_Total_eaten2)
```
```{r}
LM_Total_eaten3 <- update(LM_Total_eaten2, . ~ . - Treatment_bis:Species)
drop1(LM_Total_eaten3)
```
```{r}
LM_Total_eaten4 <- update(LM_Total_eaten3, . ~ . - Sex:Species)
drop1(LM_Total_eaten4)
```
On arrête la simplification de modèle.

Est-ce qu'elle a été utile ?
```{r}
AIC(LM_Total_eaten4)
AIC(LM_Total_eaten)
```
Oui, elle a été utile.
```{r}
LM_Total_eaten <- LM_Total_eaten4
```

### Treatment VS Treatment_bis

```{r}
AIC(lmer(log(Total_eaten) ~ Treatment_bis + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data))
AIC(lmer(log(Total_eaten) ~ Treatment + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment:Region, data = Data))
```
```{r}
summary(lmer(log(Total_eaten) ~ Treatment_bis + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data))$coeff
summary(lmer(log(Total_eaten) ~ Treatment + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment:Region, data = Data))$coeff
```

Au vu des summary et de l'AIC je préfère le modèle avec Treatment_bis

Analyse des P-val
```{r}
summary(lmer(log(Total_eaten) ~ Treatment_bis + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data))
summary(lmer(log(Total_eaten) ~ Treatment_bis + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_ref_full))
summary(lmer(log(Total_eaten) ~ Treatment_bis + Sex + Species + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_ref_smart))
```

```{r}
Info_Total_eaten <- "
Aucun traitement n'a d'effet significatif.
Les chenilles mangent significativement plus dans le traitement Mixt_smart quand elles viennent du luxembourg que quand elles viennent du brabant.
L'espèce OP mange moins que l'espèce AE
Les mangent presquent significativement (P = 0.0659) moins que les femelles.
"
Info_Total_eaten
```


### Visualisation
```{r, message=FALSE}
Summary_Total_eaten <- Data %>% 
  filter(Sex != "NA") %>% 
  group_by(Treatment_bis, Species, Sex) %>% 
  summarise(Mean=mean(log(Total_eaten)),
            Var=var(log(Total_eaten)),
            N=n()) %>% 
  mutate(CI_S=Mean+qt(0.975, N-1)*sqrt(Var/N),
         CI_I=Mean-qt(0.975, N-1)*sqrt(Var/N))
```


```{r}
ggplot(Summary_Total_eaten, aes(x=Treatment_bis, y=Mean, color = Species))+
  geom_errorbar(aes(ymin=CI_I, ymax=CI_S),width=0.2,size = 1.2, alpha = 0.5) +
  geom_point(alpha = 0.8) +
  facet_wrap(~Sex) +
  labs(title = "Quantité mangée", x = "Traitement", y = "log(Quantitée mangée (morceaux de feuilles))", color = "Species") +
  theme_bw()
```
```{r, message=FALSE}
Summary_Total_eaten2 <- Data %>% 
  filter(Sex != "NA") %>% 
  group_by(Treatment_bis, Species, Region) %>% 
  summarise(Mean=mean(log(Total_eaten)),
            Var=var(log(Total_eaten)),
            N=n()) %>% 
  mutate(CI_S=Mean+qt(0.975, N-1)*sqrt(Var/N),
         CI_I=Mean-qt(0.975, N-1)*sqrt(Var/N))
```


```{r}
ggplot(Summary_Total_eaten2, aes(x=Treatment_bis, y=Mean, color = Species))+
  geom_errorbar(aes(ymin=CI_I, ymax=CI_S),width=0.2,size = 1.2, alpha = 0.5) +
  geom_point(alpha = 0.8) +
  facet_wrap(~Region) +
  labs(title = "Total Eaten", x = "Treatment", y = "log(Total eaten)") +
  theme_bw()
```




## Pupa_mass * TT_pupation


### Test
```{r}
LM_Pupa_mass_by_TT_pupation <- lmer(Pupa_mass ~ TT_pupation + Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex + Treatment_bis:Region + TT_pupation:Species + TT_pupation:Sex + TT_pupation:Region + TT_pupation:Treatment_bis, data = Data)
summary(LM_Pupa_mass_by_TT_pupation)
```

Test des hypothèses

Normalité des réisuds
```{r}
qqnorm(resid(LM_Pupa_mass_by_TT_pupation))
qqline(resid(LM_Pupa_mass_by_TT_pupation))
```
On doit modifier les données (comme dans Pupa_mass en fait)

```{r}
LM_Pupa_mass_by_TT_pupation <- lmer(log(Pupa_mass) ~ TT_pupation + Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex + Treatment_bis:Region + TT_pupation:Species + TT_pupation:Sex + TT_pupation:Region + TT_pupation:Treatment_bis, data = Data[-239,])
# Honnêtement sans trop savoir pourquoi, la ligne 239 possédait des distance de cook très élevée par rapport au reste des points, nous l'avons donc retirée) 
summary(LM_Pupa_mass_by_TT_pupation)
```

```{r}
qqnorm(resid(LM_Pupa_mass_by_TT_pupation))
qqline(resid(LM_Pupa_mass_by_TT_pupation))
```



Homogénité des variances
```{r}
plot(LM_Pupa_mass_by_TT_pupation)
```

Distance de cook
```{r}
plot(cooks.distance(LM_Pupa_mass_by_TT_pupation))
```

### Simplification du modèle

```{r}
drop1(LM_Pupa_mass_by_TT_pupation)
```
```{r}
LM_Pupa_mass_by_TT_pupation1 <- update(LM_Pupa_mass_by_TT_pupation, . ~ . - Treatment_bis:Species)
drop1(LM_Pupa_mass_by_TT_pupation1)
```
```{r}
LM_Pupa_mass_by_TT_pupation2 <- update(LM_Pupa_mass_by_TT_pupation1, . ~ . - Sex:Species)
drop1(LM_Pupa_mass_by_TT_pupation2)
```
```{r}
LM_Pupa_mass_by_TT_pupation3 <- update(LM_Pupa_mass_by_TT_pupation2, . ~ . - TT_pupation:Treatment_bis)
drop1(LM_Pupa_mass_by_TT_pupation3)
```
```{r}
LM_Pupa_mass_by_TT_pupation4 <- update(LM_Pupa_mass_by_TT_pupation3, . ~ . - Treatment_bis:Region)
drop1(LM_Pupa_mass_by_TT_pupation4)
```
```{r}
LM_Pupa_mass_by_TT_pupation5 <- update(LM_Pupa_mass_by_TT_pupation4, . ~ . - TT_pupation:Species)
drop1(LM_Pupa_mass_by_TT_pupation5)
```
```{r}
LM_Pupa_mass_by_TT_pupation6 <- update(LM_Pupa_mass_by_TT_pupation5, . ~ . - TT_pupation:Sex)
drop1(LM_Pupa_mass_by_TT_pupation6)
```

```{r}
LM_Pupa_mass_by_TT_pupation7 <- update(LM_Pupa_mass_by_TT_pupation6, . ~ . - Treatment_bis:Sex)
drop1(LM_Pupa_mass_by_TT_pupation7)
```


Il n'y a plus d'interactions qui n'ont pas de P-val proche de 0.05, nous décidons d'arrêter la simplification du modèle.

Est ce que cela a été utile
```{r}
AIC(LM_Pupa_mass_by_TT_pupation7)
AIC(LM_Pupa_mass_by_TT_pupation)
```
Oui ca a été utile
```{r}
LM_Pupa_mass_by_TT_pupation <- LM_Pupa_mass_by_TT_pupation7
```


### Treatment VS Teatment_bis

```{r}
AIC(lmer(log(Pupa_mass) ~ TT_pupation + Treatment_bis + Sex + Species +  Region + (1 | Mother) + (1 | Site) + (1 | Room) + Species:Region +  TT_pupation:Region, data = Data[-239, ]))
AIC(lmer(log(Pupa_mass) ~ TT_pupation + Treatment + Sex + Species +  Region + (1 | Mother) + (1 | Site) + (1 | Room) + Species:Region +  TT_pupation:Region, data = Data[-239, ]))
```


```{r}
summary(lmer(log(Pupa_mass) ~ TT_pupation + Treatment_bis + Sex + Species +  Region + (1 | Mother) + (1 | Site) + (1 | Room) + Species:Region +  TT_pupation:Region, data = Data[-239, ]))$coeff
summary(lmer(log(Pupa_mass) ~ TT_pupation + Treatment + Sex + Species +  Region + (1 | Mother) + (1 | Site) + (1 | Room) + Species:Region +  TT_pupation:Region, data = Data[-239, ]))$coeff
```

L'AIC suggère de prendre Treatment mais les p-val disent que Mixt_Smart a un effet fort significatif Je vais rester avec Treatment_bis


Analyse des pval
```{r}
summary(lmer(log(Pupa_mass) ~ TT_pupation + Treatment_bis + Sex + Species +  Region + (1 | Mother) + (1 | Site) + (1 | Room) + Species:Region +  TT_pupation:Region, data = Data[-239, ]))
summary(lmer(log(Pupa_mass) ~ TT_pupation + Treatment_bis + Sex + Species +  Region + (1 | Mother) + (1 | Site) + (1 | Room) + Species:Region +  TT_pupation:Region, data = Data_ref_full[-239, ]))
summary(lmer(log(Pupa_mass) ~ TT_pupation + Treatment_bis + Sex + Species +  Region + (1 | Mother) + (1 | Site) + (1 | Room) + Species:Region +  TT_pupation:Region, data = Data_ref_smart[-239, ]))
```

```{r} 
Info_Pupa_mass_by_TT_pupation <- "
Au plus le temps pour entrer en chrysalide est élevé, au plus la chrysalide sera lourde.
ordre de croissance de masse : Dark < Full et Smart (les deux sont très simillaires) < Mixt_smart
OP est plus légère
Les chrysalides provenants de la région Luxembourg sont quasiment significativement (pval = 0.065) plus lourde.
L'impacte de la région Luxembourg est presque significativement différent (pval = 0.068) en fonction de l'espèce (OP plus légère)
L'impacte du temps avant d'atteindre la chrysalide est quasiment significativement différent fonction de la région d'ou vient la chenille (luxembourgeoise plus légère pour un même temps d'atteinte de chrysalide)
"
Info_Pupa_mass_by_TT_pupation
```

### Visualisation

```{r}
ggplot(data = Data[-239,], aes(x = TT_pupation, y = Pupa_mass)) +
  geom_point(aes(color = Treatment_bis, shape = Species)) +
  geom_smooth(method = "lm", aes(linetype = Species)) +
  labs(title = "Masse des chrysalides en fonction du temps passé pour entrer en chrysalide", x = "Temps pour entrer en chrysalide (jours)", y = "Poids des chrysalides (g)", color = "Traitement", linetype = "Espèce", shape = "Espèce") +
  theme_bw()
```

```{r}
ggplot(data = Data[-239,], aes(x = TT_pupation, y = log(Pupa_mass))) +
  geom_point(aes(color = Treatment_bis, shape = Species)) +
  geom_smooth(method = "lm", aes(color= Treatment_bis, linetype = Species)) +
  facet_wrap(~Region) +
  labs(title = "Pupa Mass by Time to pupa", x = "Time to pupation", y = "log(Pupa_mass)") +
  theme_bw()
```


## Mass_by_Time

Dévelloppement du jeux de données
```{r}
Data_mass <- Data[,1:31] %>% pivot_longer(cols = M0:M20, names_to = "Numéro_de_masse", values_to = "Mass")
Data_mass$Time <- rep(0:20, length.out = nrow(Data_mass))
```

```{r}
Data_mass_ref_smart <- Data_mass
Data_mass_ref_smart$Treatment <- relevel(Data_mass$Treatment, ref = "Smart")
Data_mass_ref_smart$Treatment_bis <- relevel(Data_mass$Treatment_bis, ref = "Smart")

Data_mass_ref_full <- Data_mass
Data_mass_ref_full$Treatment <- relevel(Data_mass$Treatment, ref = "Full")
Data_mass_ref_full$Treatment_bis <- relevel(Data_mass$Treatment_bis, ref = "Full")
```


### Test

```{r}
LM_mass_by_time <- lmer(Mass ~ Time + Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex + Treatment_bis:Region + Time:Species + Time:Sex + Time:Region + Time:Treatment_bis, data = Data_mass)
summary(LM_mass_by_time)
```

Vérification des hypothèses

```{r}
qqnorm(resid(LM_mass_by_time))
qqline(resid(LM_mass_by_time))
```
Il faudrait modifier les données

```{r}
LM_mass_by_time <- lmer(log(Mass) ~ sqrt(Time) + Treatment_bis + Sex + Species + Region + (1|Mother) + (1|Site) + (1|Room) + Sex:Species + Region:Species + Treatment_bis:Species + Treatment_bis:Sex + Treatment_bis:Region + Time:Species + Time:Sex + Time:Region + Time:Treatment_bis, data = Data_mass)
#summary(LM_mass_by_time)
```

```{r}
qqnorm(resid(LM_mass_by_time))
qqline(resid(LM_mass_by_time))
```



Homogénité des variances
```{r}
plot(LM_mass_by_time)
```

Distance de cook
```{r}
plot(cooks.distance(LM_mass_by_time))
```
On pourrait retirer quelques outlier mais cela semle bon

### Simplification du modèle

```{r}
drop1(LM_mass_by_time)
```
```{r}
LM_mass_by_time1 <- update(LM_mass_by_time, . ~ . - Species:Region)
drop1(LM_mass_by_time1)
```
```{r}
LM_mass_by_time2 <- update(LM_mass_by_time1, . ~ . - Region:Time)
drop1(LM_mass_by_time2)
```
```{r}
LM_mass_by_time3 <- update(LM_mass_by_time2, . ~ . - Treatment_bis:Sex)
drop1(LM_mass_by_time3)
```
```{r}
LM_mass_by_time4 <- update(LM_mass_by_time3, . ~ . - Sex:Species)
drop1(LM_mass_by_time4)
```
```{r}
LM_mass_by_time5 <- update(LM_mass_by_time4, . ~ . - Sex:Time)
drop1(LM_mass_by_time5)
```
```{r}
LM_mass_by_time6 <- update(LM_mass_by_time5, . ~ . - Treatment_bis:Species)
drop1(LM_mass_by_time6)
```
Plus aucune interaction ne peut être retirée, on arrête la simplification ici.

A-t-elle été utile ?
```{r}
AIC(LM_mass_by_time6)
AIC(LM_mass_by_time)
```
Oui

```{r}
LM_mass_by_time <- LM_mass_by_time6
```


### Treatment VS Teatment_bis

```{r}
AIC(lmer(log(Mass) ~ sqrt(Time) + Treatment_bis + Sex + Species + Region +  (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Species:Time + Treatment_bis:Time, data = Data_mass))
AIC(lmer(log(Mass) ~ sqrt(Time) + Treatment + Sex + Species + Region +  (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Species:Time + Treatment:Time, data = Data_mass))
```
```{r}
summary(lmer(log(Mass) ~ sqrt(Time) + Treatment_bis + Sex + Species + Region +  (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Species:Time + Treatment_bis:Time, data = Data_mass))
summary(lmer(log(Mass) ~ sqrt(Time) + Treatment + Sex + Species + Region +  (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Species:Time + Treatment:Time, data = Data_mass))
```
On a des p-val plus intéressentes avec le treatment mais treatment_bis montre des effets différents pour smart et mixt_smart. Si je me base sur les AIC je continue avec treatment_bis

Analyse des P-val
```{r}
summary(lmer(log(Mass) ~ sqrt(Time) + Treatment_bis + Sex + Species + Region +  (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Species:Time + Treatment_bis:Time, data = Data_mass))
summary(lmer(log(Mass) ~ sqrt(Time) + Treatment_bis + Sex + Species + Region +  (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Species:Time + Treatment_bis:Time, data = Data_mass_ref_full))
summary(lmer(log(Mass) ~ sqrt(Time) + Treatment_bis + Sex + Species + Region +  (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Species:Time + Treatment_bis:Time, data = Data_mass_ref_smart))
```


```{r}
Info_Mass_by_Time <- "
temps a un effet significatif positif
Pour un même temps, Les smart sont significativements plus légère que les Dark
Pour un même temps, l'Espèce OP est plus lourde.
Pour un même temps, les individus de la région Luxembourg sont moins lourds
Les individus Smart réagissent différement en fonction de si ils viennent du Luxembourg ou pas (si ils viennent du Luxembourg ils sont plus lourds pour un même temps).
Idem pour Mixt_smart mais ils sont moinds lourds
L'espèce AE réagit différement que OP à l'augmentation de temps (les p val)
"
Info_Mass_by_Time
```


### Visualisation

```{r}
ggplot(data = Data_mass, aes(x = log(Time), y = log(Mass))) +
  geom_jitter(aes(color = Treatment_bis, shape = Region), alpha = 0.33) +
  geom_smooth(method = "lm", aes(linetype = Region)) +
  facet_wrap(~Species) +
  labs(title = "Poids des individus en fonction du temps", x = "Temps (log(Numéro de mesure))", y = "log(Poids (g))", color = "Traitement", linetype = "Région", shape = "Région") +
  theme_bw()
```

```{r}
ggplot(data = Data_mass, aes(x = sqrt(Time), y = log(Mass))) +
  geom_point(aes(color = Treatment_bis, shape = Species), alpha = 0.5) +
  geom_smooth(method = "lm", aes(color= Treatment_bis, linetype = Species)) +
  facet_wrap(~Region) +
  labs(title = "Mass by Time", x = "Time (sqrt(Weeks)", y = "log(Mass)") +
  theme_bw()
```






# Analyses Lumière

## Jeux de donnée avec références


```{r}
Data_refFTL_Night <- Data
Data_refFTL_Night$Treatment_FTL <- relevel(Data$Treatment_FTL, ref = "Night")
Data_refFTL_Night$Treatment_FTL <- relevel(Data$Treatment_FTL, ref = "Night")

Data_refFTL_Mix <- Data
Data_refFTL_Mix$Treatment_FTL <- relevel(Data$Treatment_FTL, ref = "Mix")
Data_refFTL_Mix$Treatment_FTL <- relevel(Data$Treatment_FTL, ref = "Mix")
```

## Did_move

### Test

```{r,warning=FALSE, message=FALSE}
GLM_Did_move <- glmer(Did_move ~ Treatment_bis + Sex  + Region + Treatment_FTL + Light_orientation + (1|Mother) + (1|Site) + (1|Room)  + Treatment_bis:Sex + Treatment_bis:Region + Treatment_FTL:Sex + Treatment_FTL:Region + Treatment_FTL:Treatment_bis + Light_orientation:Sex + Light_orientation:Region + Light_orientation:Treatment_bis, family = binomial, data = Data)
summary(GLM_Did_move)
```


### Simplification du modèle

Simplification par AIC/BIC
```{r, warning=FALSE, message=FALSE}
drop1(GLM_Did_move)
```
```{r,warning=FALSE, message=FALSE}
GLM_Did_move2 <- update(GLM_Did_move, . ~ . - Treatment_bis:Treatment_FTL)
drop1(GLM_Did_move2)
```
```{r,warning=FALSE, message=FALSE}
GLM_Did_move3 <- update(GLM_Did_move2, . ~ . - Treatment_bis:Sex)
drop1(GLM_Did_move3)
```
```{r,warning=FALSE, message=FALSE}
GLM_Did_move4 <- update(GLM_Did_move3, . ~ . - Sex:Treatment_FTL)
drop1(GLM_Did_move4)
```
```{r,warning=FALSE, message=FALSE}
GLM_Did_move5 <- update(GLM_Did_move4, . ~ . - Treatment_bis:Light_orientation)
drop1(GLM_Did_move5)
```
Est ce qu'il faut pas juste retirer toutes les interactions ?

```{r,warning=FALSE, message=FALSE}
BIC(glmer(Did_move ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data, family = binomial))
BIC(GLM_Did_move5)
BIC(GLM_Did_move4)
BIC(GLM_Did_move3)
BIC(GLM_Did_move2)
BIC(GLM_Did_move)
```

Si

```{r, warning=FALSE, message=FALSE}
GLM_Did_move <- glmer(Did_move ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data, family = binomial)
```

### Treatment VS Treatment_bis

```{r, warning=FALSE, message=FALSE}
AIC(glmer(Did_move ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data, family = binomial))
AIC(glmer(Did_move ~ Treatment + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data, family = binomial))
```

```{r, warning=FALSE, message=FALSE}
summary(glmer(Did_move ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data, family = binomial))
summary(glmer(Did_move ~ Treatment + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data, family = binomial))
```

Aucune grosse différence mais vu qu'on a utilisé treatment_bis dans les autres tests, nous allons continuer avec treatment.

### Résultats

Analyse des p-val

```{r}
summary(glmer(Did_move ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data, family = binomial))
summary(glmer(Did_move ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data_ref_full, family = binomial))
summary(glmer(Did_move ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data_ref_smart, family = binomial))
summary(glmer(Did_move ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data_ref_mixt_smart, family = binomial))
summary(glmer(Did_move ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data_refFTL_Mix, family = binomial))
summary(glmer(Did_move ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data_refFTL_Night, family = binomial))
```

```{r}
test1 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Day", as.data.frame(( summary(glmer(Did_move ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data, family = binomial))$coeff)[,]))

test2 <- data.frame(Reference_Treatment = "Smart", Reference_Timing = "Day", as.data.frame(( summary(glmer(Did_move ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data_ref_smart, family = binomial))$coeff)[,]))

test3 <- data.frame(Reference_Treatment = "Full", Reference_Timing = "Day", as.data.frame(( summary(glmer(Did_move ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data_ref_full, family = binomial))$coeff)[,]))

test4 <- data.frame(Reference_Treatment = "Mixt_smart", Reference_Timing = "Day", as.data.frame(( summary(glmer(Did_move ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data_ref_mixt_smart, family = binomial))$coeff)[,]))

test5 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Mixt", as.data.frame(( summary(glmer(Did_move ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data_refFTL_Mix, family = binomial))$coeff)[,]))

test6 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Night", as.data.frame(( summary(glmer(Did_move ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data_refFTL_Night, family = binomial))$coeff)[,]))



Results_Did_move <- rbind(test1,test2,test3,test4,test5,test6)

colnames(Results_Did_move ) <- c("Reference_Treatment", "Reference_Timing" ,"Estimate", "Std. Error", "z value", "Pr(>|z|)")

write.csv2( Results_Did_move , file = "output/Results_Did_move.csv" )


kable(Results_Did_move , caption = " Probabilité de bouger ", padding = 1 )
```

```{r}
Info_Did_move <- "
Aucun effet significatif
"
```

### Visualisation

```{r}
ggplot(Data %>% filter(Did_move != "NA")) +
  aes(x = Did_move, fill = Treatment_FTL) +
  geom_bar(position = "dodge") +
  scale_fill_hue(direction = 1) +
  labs(title= "Probabilité de bouger", x = "A bougé", y = "Nombre d'individus", fill = "Timing") +
  theme_minimal()
```

## Did_FTL

### Test

```{r,warning=FALSE, message=FALSE}
GLM_Did_FTL <- glmer(Did_FTL ~ Treatment_bis + Sex  + Region + Treatment_FTL + Light_orientation + (1|Mother) + (1|Site) + (1|Room)  + Treatment_bis:Sex + Treatment_bis:Region + Treatment_FTL:Sex + Treatment_FTL:Region + Treatment_FTL:Treatment_bis + Light_orientation:Sex + Light_orientation:Region + Light_orientation:Treatment_bis, family = binomial, data = Data)
summary(GLM_Did_FTL)
```


### Simplification du modèle

Simplification par AIC/BIC
```{r, warning=FALSE, message=FALSE}
drop1(GLM_Did_FTL)
```
```{r,warning=FALSE, message=FALSE}
GLM_Did_FTL2 <- update(GLM_Did_FTL, . ~ . - Treatment_bis:Sex)
drop1(GLM_Did_FTL2)
```
```{r,warning=FALSE, message=FALSE}
GLM_Did_FTL3 <- update(GLM_Did_FTL2, . ~ . - Sex:Treatment_FTL)
drop1(GLM_Did_FTL3)
```
```{r,warning=FALSE, message=FALSE}
GLM_Did_FTL4 <- update(GLM_Did_FTL3, . ~ . - Treatment_bis:Treatment_FTL)
drop1(GLM_Did_FTL4)
```
```{r,warning=FALSE, message=FALSE}
GLM_Did_FTL5 <- update(GLM_Did_FTL4, . ~ . - Region:Light_orientation)
drop1(GLM_Did_FTL5)
```
```{r,warning=FALSE, message=FALSE}
GLM_Did_FTL6 <- update(GLM_Did_FTL5, . ~ . - Treatment_bis:Light_orientation)
drop1(GLM_Did_FTL6)
```
```{r,warning=FALSE, message=FALSE}
GLM_Did_FTL7 <- update(GLM_Did_FTL6, . ~ . - Region:Treatment_FTL)
drop1(GLM_Did_FTL7)
```

Est ce qu'il faut pas juste retirer toutes les interactions ?

```{r,warning=FALSE, message=FALSE}
BIC(glmer(Did_FTL ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data, family = binomial))
BIC(GLM_Did_FTL7)
```
Si

```{r, warning=FALSE, message=FALSE}
GLM_Did_FTL <- glmer(Did_FTL ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data, family = binomial)
```

### Treatment VS Treatment_bis

```{r, warning=FALSE, message=FALSE}
AIC(glmer(Did_FTL ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data, family = binomial))
AIC(glmer(Did_FTL ~ Treatment + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data, family = binomial))
```

```{r, warning=FALSE, message=FALSE}
summary(glmer(Did_FTL ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data, family = binomial))
summary(glmer(Did_FTL ~ Treatment + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data, family = binomial))
```

Il faudrait muieux continuer avec Treatment

### Résultats

Analyse des p-val

```{r}
summary(glmer(Did_FTL ~ Treatment + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data, family = binomial))
summary(glmer(Did_FTL ~ Treatment + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data_ref_full, family = binomial))
summary(glmer(Did_FTL ~ Treatment + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data_ref_smart, family = binomial))
summary(glmer(Did_FTL ~ Treatment + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data_refFTL_Mix, family = binomial))
summary(glmer(Did_FTL ~ Treatment + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data_refFTL_Night, family = binomial))
```

```{r}
test1 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Day", as.data.frame(( summary(glmer(Did_FTL ~ Treatment + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data, family = binomial))$coeff)[,]))

test2 <- data.frame(Reference_Treatment = "Smart", Reference_Timing = "Day", as.data.frame(( summary(glmer(Did_FTL ~ Treatment + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data_ref_smart, family = binomial))$coeff)[,]))

test3 <- data.frame(Reference_Treatment = "Full", Reference_Timing = "Day", as.data.frame(( summary(glmer(Did_FTL ~ Treatment + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data_ref_full, family = binomial))$coeff)[,]))

test4 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Mixt", as.data.frame(( summary(glmer(Did_FTL ~ Treatment + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data_refFTL_Mix, family = binomial))$coeff)[,]))

test5 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Night", as.data.frame(( summary(glmer(Did_FTL ~ Treatment + Sex + Region + Treatment_FTL + Light_orientation +  (1 | Mother) + (1 | Site) + (1 | Room), data = Data_refFTL_Night, family = binomial))$coeff)[,]))



Results_Did_FTL <- rbind(test1,test2,test3,test4,test5)

colnames(Results_Did_FTL ) <- c("Reference_Treatment", "Reference_Timing" ,"Estimate", "Std. Error", "z value", "Pr(>|z|)")

write.csv2( Results_Did_FTL , file = "output/Results_Did_FTL.csv" )

kable(Results_Did_FTL , caption = " Probabilité de voler à la lumière ", padding = 1 )
```

```{r}
Info_Did_FTL <- "
Les individus expérimentés durant la nuit volent significativement plus que ceux expérimentés durant le jour. Les individus expérimentés avec la méthode (timing de l'expérience) Mixte ne sont pas significativement différent ni de ceux expérimentés pendant la nuit ni de ceux expérimentés pendant le jour.
"
```




### Visualisation

```{r}
ggplot(Data %>% filter(Did_FTL != "NA")) +
  aes(x = Did_FTL, fill = Treatment_FTL) +
  geom_bar(position = "dodge") +
  scale_fill_hue(direction = 1) +
  labs(title= "Probabilité de voler à la lumière", x = "A volé à la lumière", y = "Nombre d'individus", fill = "Timing") +
  theme_minimal()
```




## First_move

### Test

```{r}
LM_First_move <- lmer(First_move ~ Treatment_bis + Sex  + Region + Treatment_FTL + Light_orientation + (1|Mother) + (1|Site) + (1|Room)  + Treatment_bis:Sex + Treatment_bis:Region + Treatment_FTL:Sex + Treatment_FTL:Region + Treatment_FTL:Treatment_bis + Light_orientation:Sex + Light_orientation:Region + Light_orientation:Treatment_bis, data = Data)
summary(LM_First_move)
```

Vérification des hypothèses

Normalité
```{r}
qqnorm(resid(LM_First_move))
qqline(resid(LM_First_move))
```
Il faudrait modifier les données




```{r}
LM_First_move <- lmer(log(First_move+0.1) ~ Treatment_bis + Sex  + Region + Treatment_FTL + Light_orientation + (1|Mother) + (1|Site) + (1|Room)  + Treatment_bis:Sex + Treatment_bis:Region + Treatment_FTL:Sex + Treatment_FTL:Region + Treatment_FTL:Treatment_bis + Light_orientation:Sex + Light_orientation:Region + Light_orientation:Treatment_bis, data = Data )
summary(LM_First_move)
```

```{r}
qqnorm(resid(LM_First_move))
qqline(resid(LM_First_move))
```

Homogénité des variances
```{r}
plot(LM_First_move)
```


Distance de cook
```{r}
plot(cooks.distance(LM_First_move))
```


### Simplification du modèle

```{r}
drop1(LM_First_move)
```
```{r}
LM_First_move1 <- update(LM_First_move, . ~ . - Region:Treatment_FTL)
drop1(LM_First_move1)
```
```{r}
LM_First_move2 <- update(LM_First_move1, . ~ . - Sex:Light_orientation)
drop1(LM_First_move2)
```
```{r}
LM_First_move3 <- update(LM_First_move2, . ~ . - Treatment_bis:Light_orientation)
drop1(LM_First_move3)
```
```{r}
LM_First_move4 <- update(LM_First_move3, . ~ . - Treatment_bis:Treatment_FTL)
drop1(LM_First_move4)
```
```{r}
LM_First_move5 <- update(LM_First_move4, . ~ . - Treatment_bis:Sex)
drop1(LM_First_move5)
```
```{r}
LM_First_move6 <- update(LM_First_move5, . ~ . - Region:Light_orientation)
drop1(LM_First_move6)
```
```{r}
LM_First_move7 <- update(LM_First_move6, . ~ . - Sex:Treatment_FTL)
drop1(LM_First_move7)
```


Le modèle ne sait pas être plus simplifié. Au vu de la complexité du modèle initial (nombre de paramètres), nous choisissons d'utiliser le BIC et non le AIC.

```{r}
BIC(LM_First_move7)
BIC(LM_First_move)
```
La simplification fu utile

```{r}
LM_First_move <- LM_First_move7
```

### Treatment VS Teatment_bis

```{r}
AIC(lmer(log(First_move+0.1) ~ Treatment_bis + Region + Treatment_FTL + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region  , data = Data))
AIC(lmer(log(First_move+0.1) ~ Treatment + Region + Treatment_FTL + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment:Region + Treatment:Treatment_FTL, data = Data))
```


```{r}
summary(lmer(log(First_move+0.1) ~ Treatment_bis + Region + Treatment_FTL + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region  , data = Data))$coeff
summary(lmer(log(First_move+0.1) ~ Treatment + Region + Treatment_FTL + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment:Region + Treatment:Treatment_FTL, data = Data))$coeff
```

Le modèle avec treatment_bis est mieux

### Résultats

Analyse des pval

```{r}
summary(lmer(log(First_move+0.1) ~ Treatment_bis + Region + Treatment_FTL + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region  , data = Data))
summary(lmer(log(First_move+0.1) ~ Treatment_bis + Region + Treatment_FTL + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region  , data = Data_ref_full))
summary(lmer(log(First_move+0.1) ~ Treatment_bis + Region + Treatment_FTL + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region  , data = Data_ref_smart))
summary(lmer(log(First_move+0.1) ~ Treatment_bis + Region + Treatment_FTL + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region  , data = Data_ref_mixt_smart))
summary(lmer(log(First_move+0.1) ~ Treatment_bis + Region + Treatment_FTL + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region  , data = Data_refFTL_Night))
summary(lmer(log(First_move+0.1) ~ Treatment_bis + Region + Treatment_FTL + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region  , data = Data_refFTL_Mix))
```

```{r}
test1 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(First_move+0.1) ~ Treatment_bis + Region + Treatment_FTL + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region  , data = Data))$coeff)[,]))

test2 <- data.frame(Reference_Treatment = "Smart", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(First_move+0.1) ~ Treatment_bis + Region + Treatment_FTL + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region  , data = Data_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference_Treatment = "Full", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(First_move+0.1) ~ Treatment_bis + Region + Treatment_FTL + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region  , data = Data_ref_full))$coeff)[,]))

test4 <- data.frame(Reference_Treatment = "Mixt_smart", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(First_move+0.1) ~ Treatment_bis + Region + Treatment_FTL + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region  , data = Data_ref_mixt_smart))$coeff)[,]))

test5 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Mixt", as.data.frame(( summary(lmer(log(First_move+0.1) ~ Treatment_bis + Region + Treatment_FTL + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region  , data = Data_refFTL_Mix))$coeff)[,]))

test6 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Night", as.data.frame(( summary(lmer(log(First_move+0.1) ~ Treatment_bis + Region + Treatment_FTL + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region  , data = Data_refFTL_Night))$coeff)[,]))



Results_First_move <- rbind(test1,test2,test3,test4,test5,test6)

colnames(Results_First_move ) <- c("Reference_Treatment", "Reference_Timing" ,"Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_First_move , file = "output/Results_First_move.csv" )

kable(Results_First_move , caption = " Temps mis pour faire le premier mouvement ", padding = 1 )
```

```{r}
Info_First_move <- "
Le traitement Night diminue le temps mis avant le premier mouvement. Day et Mix ne sont pas significativement différent.
L'impacte du traitement Mixt_smart est différent (plus rapide dans le luxembourg) en fonction de la région d'ou provient la mère de la chenille qui passe le test. (Il est toujours significatif sauf quand comparé avec Full mais là la p val est de 0.06634)
"
Info_First_move
```


### Visualisation

```{r, message=FALSE}
Summary_First_move <- Data %>% 
  filter(Treatment_FTL != "NA") %>% 
  group_by(Region, Treatment_FTL, Treatment_bis) %>% 
  summarise(Mean=mean(log(First_move+0.1)),
            Var=var(log(First_move+0.1)),
            N=n()) %>% 
  mutate(CI_S=Mean+qt(0.975, N-1)*sqrt(Var/N),
         CI_I=Mean-qt(0.975, N-1)*sqrt(Var/N))
```


```{r}
ggplot(Summary_First_move, aes(x=Treatment_bis, y=Mean, color = Treatment_FTL))+
  geom_errorbar(aes(ymin=CI_I, ymax=CI_S),width=0.2,size = 1.2, alpha = 0.5, position = "dodge") +
  geom_point(alpha = 1) +
  facet_wrap(~Region) +
  labs(title = "First move", x = "Treatment", y = "log(Time to have the first move)") +
  theme_bw()
```

```{r}
ggplot(data = Data %>% filter(Treatment_FTL != "NA"), aes(x = Treatment_bis, y = log(First_move+0.1), color = Treatment_FTL)) +
  geom_boxplot() +
  geom_point(position = position_dodge(width = 1), alpha = 0.75, data = Data %>% filter(Treatment_FTL == "Day" | Treatment_FTL == "Night")) +
  geom_point(alpha = 0.75, data = Data %>% filter(Treatment_FTL == "Mix")) +
  labs(title = "Temps pour faire le premier mouvement", x = "Tratiement", y = "log(Temps pour faire le premier mouvement (s))", color = "Timing" ) +
  theme_bw()
```


## Flight_to_light

### Test


```{r}
LM_Flight_to_light <- lmer(Flight_to_light ~ Treatment_bis + Sex  + Region + Treatment_FTL + Light_orientation + (1|Mother) + (1|Site) + (1|Room)  + Treatment_bis:Sex + Treatment_bis:Region + Treatment_FTL:Sex + Treatment_FTL:Region + Treatment_FTL:Treatment_bis + Light_orientation:Sex + Light_orientation:Region + Light_orientation:Treatment_bis, data = Data)
summary(LM_Flight_to_light)
```
Normalité des résidus
```{r}
qqnorm(resid(LM_Flight_to_light))
qqline(resid(LM_Flight_to_light))
```



Homogénité des variances
```{r}
plot(LM_Flight_to_light)
```
légère forme d'entonnoir mais c'est passable

Distance de cook
```{r}
plot(cooks.distance(LM_Flight_to_light))
```

```{r}
(cooks.distance(LM_Flight_to_light))
```

Nous avons un problème avec le point de la ligne 252 sans trop savoir pourquoi honnètement. On va juste le retirer

```{r}
LM_Flight_to_light <- lmer(Flight_to_light ~ Treatment_bis + Sex  + Region + Treatment_FTL + Light_orientation + (1|Mother) + (1|Site) + (1|Room)  + Treatment_bis:Sex + Treatment_bis:Region + Treatment_FTL:Sex + Treatment_FTL:Region + Treatment_FTL:Treatment_bis + Light_orientation:Sex + Light_orientation:Region + Light_orientation:Treatment_bis, data = Data[-252,])
summary(LM_Flight_to_light)
```
Normalité des résidus
```{r}
qqnorm(resid(LM_Flight_to_light))
qqline(resid(LM_Flight_to_light))
```



Homogénité des variances
```{r}
plot(LM_Flight_to_light)
```
légère forme d'entonnoir mais c'est passable

Distance de cook
```{r}
plot(cooks.distance(LM_Flight_to_light))
```

### Simplification du modèle


```{r}
drop1(LM_Flight_to_light)
```
```{r}
LM_Flight_to_light1 <- update(LM_Flight_to_light, . ~ . - Treatment_bis:Treatment_FTL)
drop1(LM_Flight_to_light1)
```
```{r}
LM_Flight_to_light2 <- update(LM_Flight_to_light1, . ~ . - Sex:Light_orientation)
drop1(LM_Flight_to_light2)
```
```{r}
LM_Flight_to_light3 <- update(LM_Flight_to_light2, . ~ . - Treatment_bis:Light_orientation)
drop1(LM_Flight_to_light3)
```
```{r}
LM_Flight_to_light4 <- update(LM_Flight_to_light3, . ~ . - Region:Treatment_FTL)
drop1(LM_Flight_to_light4)
```
```{r}
LM_Flight_to_light5 <- update(LM_Flight_to_light4, . ~ . - Sex:Treatment_FTL)
drop1(LM_Flight_to_light5)
```
```{r}
LM_Flight_to_light6 <- update(LM_Flight_to_light5, . ~ . - Treatment_bis:Sex)
drop1(LM_Flight_to_light6)
```

Les seules interactions qui restent sont significatives donc on les laisse.


```{r}
LM_Flight_to_light <- LM_Flight_to_light6
```




### Treatment VS Teatment_bis

```{r}
AIC(lmer(log(Flight_to_light+0.1) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Region:Light_orientation , data = Data))
AIC(lmer(log(Flight_to_light+0.1) ~ Treatment + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment:Region +  Region:Light_orientation , data = Data))
```
```{r}
summary(lmer(log(Flight_to_light+0.1) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Region:Light_orientation , data = Data))$coeff
summary(lmer(log(Flight_to_light+0.1) ~ Treatment + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment:Region +  Region:Light_orientation , data = Data))$coeff
```
On garde Treatment_bis

### Résultats

Analyse des p-val

```{r}
summary(lmer(log(Flight_to_light+0.1) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Region:Light_orientation , data = Data))
summary(lmer(log(Flight_to_light+0.1) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Region:Light_orientation , data = Data_ref_full))
summary(lmer(log(Flight_to_light+0.1) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Region:Light_orientation , data = Data_ref_smart))
summary(lmer(log(Flight_to_light+0.1) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Region:Light_orientation , data = Data_ref_mixt_smart))
summary(lmer(log(Flight_to_light+0.1) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Region:Light_orientation , data = Data_refFTL_Night))
summary(lmer(log(Flight_to_light+0.1) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Region:Light_orientation , data = Data_refFTL_Mix))
```

```{r}
test1 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(Flight_to_light+0.1) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Region:Light_orientation , data = Data))$coeff)[,]))

test2 <- data.frame(Reference_Treatment = "Smart", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(Flight_to_light+0.1) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Region:Light_orientation , data = Data_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference_Treatment = "Full", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(Flight_to_light+0.1) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Region:Light_orientation , data = Data_ref_full))$coeff)[,]))

test4 <- data.frame(Reference_Treatment = "Mixt_smart", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(Flight_to_light+0.1) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Region:Light_orientation , data = Data_ref_mixt_smart))$coeff)[,]))

test5 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Mixt", as.data.frame(( summary(lmer(log(Flight_to_light+0.1) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Region:Light_orientation , data = Data_refFTL_Mix))$coeff)[,]))

test6 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Night", as.data.frame(( summary(lmer(log(Flight_to_light+0.1) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Region:Light_orientation , data = Data_refFTL_Night))$coeff)[,]))



Results_FTL <- rbind(test1,test2,test3,test4,test5,test6)

colnames(Results_FTL ) <- c("Reference_Treatment", "Reference_Timing" ,"Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_FTL , file = "output/Results_FTL.csv" )

kable(Results_FTL , caption = " Temps mis pour voler à la lumière ", padding = 1 )
```

```{r}
Info_Flight_to_light <- "
Les individus du expérimentés la nuit mettent significativement plus de temps que le traitement jour.
Le fait que la lampe  soit orientée au Sud a impacte différent en fonction du sex expérimenté (les mâle sont plus lent)
"
Info_Flight_to_light
```


### Visualisation

```{r, message=FALSE}
Summary_Flight_to_light <- Data %>% 
  filter(Treatment_FTL != "NA") %>% 
  group_by(Treatment_FTL, Treatment_bis) %>% 
  summarise(Mean=mean(log(Flight_to_light)),
            Var=var(log(Flight_to_light)),
            N=n()) %>% 
  mutate(CI_S=Mean+qt(0.975, N-1)*sqrt(Var/N),
         CI_I=Mean-qt(0.975, N-1)*sqrt(Var/N))
```


```{r}
ggplot(Summary_Flight_to_light, aes(x=Treatment_bis, y=Mean, color = Treatment_FTL))+
  geom_errorbar(aes(ymin=CI_I, ymax=CI_S),width=0.2,size = 1.2, alpha = 0.5, position = "dodge") +
  geom_point(alpha = 1) +
  labs(title = "Fly to light", x = "Treatment", y = "log(Time to have the first move)") +
  theme_bw()
```

```{r}
ggplot(data = Data, aes(x = Treatment_bis, y = Flight_to_light, color = Treatment_FTL)) +
  geom_boxplot() +
  geom_point(position = position_dodge(width = 1), alpha = 0.75, data = Data %>% filter(Treatment_FTL == "Day" | Treatment_FTL == "Night")) +
  geom_point(alpha = 0.75, data = Data %>% filter(Treatment_FTL == "Mix")) +
  labs(title = "Temps pour voler à la lumière", x = "Tratiement", y = "Temps pour voler à la lumière (s)", color = "Timing" ) +
  theme_bw()
```



## Speed to light

### Test


```{r}
LM_Speed_to_light <- lmer(Speed_to_light ~ Treatment_bis + Sex  + Region + Treatment_FTL + Light_orientation + (1|Mother) + (1|Site) + (1|Room)  + Treatment_bis:Sex + Treatment_bis:Region + Treatment_FTL:Sex + Treatment_FTL:Region + Treatment_FTL:Treatment_bis + Light_orientation:Sex + Light_orientation:Region + Light_orientation:Treatment_bis, data = Data)
summary(LM_Speed_to_light)
```
Normalité des résidus
```{r}
qqnorm(resid(LM_Speed_to_light))
qqline(resid(LM_Speed_to_light))
```


```{r}
LM_Speed_to_light <- lmer(log(Speed_to_light+0.000001) ~ Treatment_bis + Sex  + Region + Treatment_FTL + Light_orientation + (1|Mother) + (1|Site) + (1|Room)  + Treatment_bis:Sex + Treatment_bis:Region + Treatment_FTL:Sex + Treatment_FTL:Region + Treatment_FTL:Treatment_bis + Light_orientation:Sex + Light_orientation:Region + Light_orientation:Treatment_bis, data = Data)
```
Normalité des résidus
```{r}
qqnorm(resid(LM_Speed_to_light))
qqline(resid(LM_Speed_to_light))
```

Homogénité des variances
```{r}
plot(LM_Speed_to_light)
```
La forme est due au grand nombre de 0


Distance de cook
```{r}
plot(cooks.distance(LM_Speed_to_light))
```

On a un léger outlier mais je pense que cela passe sans trop de problème vu les valeurs de cook lui étant associées

### Simplification du modèle


```{r}
drop1(LM_Speed_to_light)
```
```{r}
LM_Speed_to_light1 <- update(LM_Speed_to_light, . ~ . - Sex:Treatment_FTL)
drop1(LM_Speed_to_light1)
```
```{r}
LM_Speed_to_light2 <- update(LM_Speed_to_light1, . ~ . - Region:Light_orientation)
drop1(LM_Speed_to_light2)
```
```{r}
LM_Speed_to_light3 <- update(LM_Speed_to_light2, . ~ . - Treatment_bis:Sex)
drop1(LM_Speed_to_light3)
```
```{r}
LM_Speed_to_light4 <- update(LM_Speed_to_light3, . ~ . - Treatment_bis:Treatment_FTL)
drop1(LM_Speed_to_light4)
```
```{r}
LM_Speed_to_light5 <- update(LM_Speed_to_light4, . ~ . - Treatment_bis:Region)
drop1(LM_Speed_to_light5)
```
```{r}
LM_Speed_to_light6 <- update(LM_Speed_to_light5, . ~ . - Treatment_bis:Light_orientation)
drop1(LM_Speed_to_light6)
```
```{r}
LM_Speed_to_light7 <- update(LM_Speed_to_light6, . ~ . - Region:Treatment_FTL)
drop1(LM_Speed_to_light7)
```

Simplification utile ? A nouveau, vu la complexité du modèle de base on tetse en utilisant le BIC
```{r}
BIC(LM_Speed_to_light)
BIC(LM_Speed_to_light1)
BIC(LM_Speed_to_light2)
BIC(LM_Speed_to_light3)
BIC(LM_Speed_to_light4)
BIC(LM_Speed_to_light5)
BIC(LM_Speed_to_light6)
BIC(LM_Speed_to_light7)
```
```{r}
LM_Speed_to_light <- LM_Speed_to_light3
```


### Treatment VS Teatment_bis

```{r}
BIC(lmer(Speed_to_light ~ Treatment_bis + Sex + Region + Treatment_FTL +  Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region + Region:Treatment_FTL +  Treatment_bis:Treatment_FTL + Sex:Light_orientation + Treatment_bis:Light_orientation, data = Data))
BIC(lmer(Speed_to_light ~ Treatment + Sex + Region + Treatment_FTL +  Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment:Region + Region:Treatment_FTL +  Treatment:Treatment_FTL + Sex:Light_orientation + Treatment:Light_orientation, data = Data))
```
```{r}
summary(lmer(Speed_to_light ~ Treatment_bis + Sex + Region + Treatment_FTL +  Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region + Region:Treatment_FTL +  Treatment_bis:Treatment_FTL + Sex:Light_orientation + Treatment_bis:Light_orientation, data = Data))$coeff
summary(lmer(Speed_to_light ~ Treatment + Sex + Region + Treatment_FTL +  Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment:Region + Region:Treatment_FTL +  Treatment:Treatment_FTL + Sex:Light_orientation + Treatment:Light_orientation, data = Data))$coeff
```

Le meilleur est treatment_bis

Au final, au vu du nombre d'échantillons présent dans chaque termes d'interactions et les p-val associées à ces mêmes interactions, nous décidons de partir sur le modèle le plus simplifié. 

### Résultats

Analyse des p-val

```{r}
summary(lmer(log(Speed_to_light+0.000001) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room), data = Data))
summary(lmer(log(Speed_to_light+0.000001) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_ref_full))
summary(lmer(log(Speed_to_light+0.000001) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_ref_smart))
summary(lmer(log(Speed_to_light+0.000001) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_ref_mixt_smart))
summary(lmer(log(Speed_to_light+0.000001) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_refFTL_Night))
summary(lmer(log(Speed_to_light+0.000001) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_refFTL_Mix))
```

```{r}
test1 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(Speed_to_light+0.000001) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room), data = Data))$coeff)[,]))

test2 <- data.frame(Reference_Treatment = "Smart", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(Flight_to_light) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference_Treatment = "Full", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(Speed_to_light+0.000001) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_ref_smart))$coeff)[,]))

test4 <- data.frame(Reference_Treatment = "Mixt_smart", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(Speed_to_light+0.000001) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_ref_mixt_smart))$coeff)[,]))

test5 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Mixt", as.data.frame(( summary(lmer(log(Speed_to_light+0.000001) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_refFTL_Mix))$coeff)[,]))

test6 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Night", as.data.frame(( summary(lmer(log(Speed_to_light+0.000001) ~ Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_refFTL_Night))$coeff)[,]))



Results_Speed_to_light <- rbind(test1,test2,test3,test4,test5,test6)

colnames(Results_Speed_to_light ) <- c("Reference_Treatment", "Reference_Timing" ,"Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_Speed_to_light , file = "output/Results_Speed_to_light.csv" )

kable(Results_Speed_to_light , caption = " Temps mis pour voler à la lumière ", padding = 1 )
```

```{r}
Info_Speed_to_light <- "
Les individus expérimentés la nuits ou avec la manière mixte (30 min dans l'obscurité avant de faire l'expérience de jour) sont significativement plus rapide que ceux expérimentés le jour. Néanmoins, les individus expérimentés de la manière mixte ne sont pas significativement différent de ceux expérimentés la nuit. Nous pouvons également noter le fait que les individus du traitement Full sont quasiment significativement (p-val = 0.0744) plus rapides que ceux du traitement Dark.
"
Info_Speed_to_light
```


### Visulaisation

```{r}
ggplot(Data, aes(x=Treatment_FTL, y=log(Speed_to_light+0.000001), color = Treatment_FTL))+
  geom_boxplot(data = Data %>% filter(Treatment_FTL != "NA"), outlier.shape = NA) +
  geom_jitter(width = 0.2, height = 0.1, alpha = 0.75, aes(shape = Treatment_bis)) +
  labs(title = "Vitesse pour voler à la lumière", x = "Tratiement", y = "log(vitesse pour voler à la lumière m/min)", color = "Timing", shape = "Traitement" ) +
  theme_bw()
```




## FTL_by_First_move


### Test

```{r}
LM_FTL_by_First_move <- lmer(Flight_to_light ~ First_move + Treatment_bis + Sex  + Region + Treatment_FTL + Light_orientation + (1|Mother) + (1|Site) + (1|Room)  + Treatment_bis:Sex + Treatment_bis:Region + Treatment_FTL:Sex + Treatment_FTL:Region + Treatment_FTL:Treatment_bis + Light_orientation:Sex + Light_orientation:Region + Light_orientation:Treatment_bis, data = filter(Data))
summary(LM_FTL_by_First_move)
```

```{r}
qqnorm(resid(LM_FTL_by_First_move))
qqline(resid(LM_FTL_by_First_move))
```
Il faut modifier les variables

```{r}
LM_FTL_by_First_move <- lmer(log(Flight_to_light+0.1) ~ First_move + Treatment_bis + Sex  + Region + Treatment_FTL + Light_orientation + (1|Mother) + (1|Site) + (1|Room)  + Treatment_bis:Sex + Treatment_bis:Region + Treatment_FTL:Sex + Treatment_FTL:Region + Treatment_FTL:Treatment_bis + Light_orientation:Sex + Light_orientation:Region + Light_orientation:Treatment_bis, data = Data)
summary(LM_FTL_by_First_move)
```

```{r}
qqnorm(resid(LM_FTL_by_First_move))
qqline(resid(LM_FTL_by_First_move))
```

Peut mieux faire mais on s'en contentera

Homogénité des variances
```{r}
plot(LM_FTL_by_First_move)
```


Distance de cook
```{r}
plot(cooks.distance(LM_FTL_by_First_move))
```

### Simplification du modèle

```{r}
drop1(LM_FTL_by_First_move)
```
```{r}
LM_FTL_by_First_move1 <- update(LM_FTL_by_First_move, . ~ . - Treatment_bis:Region)
drop1(LM_FTL_by_First_move1)
```
```{r}
LM_FTL_by_First_move2 <- update(LM_FTL_by_First_move1, . ~ . - Treatment_bis:Light_orientation)
drop1(LM_FTL_by_First_move2)
```
```{r}
LM_FTL_by_First_move3 <- update(LM_FTL_by_First_move2, . ~ . - Sex:Treatment_FTL)
drop1(LM_FTL_by_First_move3)
```
```{r}
LM_FTL_by_First_move4 <- update(LM_FTL_by_First_move3, . ~ . - Sex:Light_orientation)
drop1(LM_FTL_by_First_move4)
```
```{r}
LM_FTL_by_First_move5 <- update(LM_FTL_by_First_move4, . ~ . - Treatment_bis:Sex)
drop1(LM_FTL_by_First_move5)
```
```{r}
LM_FTL_by_First_move6 <- update(LM_FTL_by_First_move5, . ~ . - Sex)
drop1(LM_FTL_by_First_move6)
```
```{r}
LM_FTL_by_First_move7 <- update(LM_FTL_by_First_move6, . ~ . - Treatment_bis:Treatment_FTL)
drop1(LM_FTL_by_First_move7)
```
```{r}
LM_FTL_by_First_move8 <- update(LM_FTL_by_First_move7, . ~ . - Region:Light_orientation)
drop1(LM_FTL_by_First_move8)
```

Au vu de la complexité du modèle de base, nous utilisons le BIC
```{r}
BIC(LM_FTL_by_First_move)
BIC(LM_FTL_by_First_move8)
BIC(LM_FTL_by_First_move7)
BIC(LM_FTL_by_First_move6)
BIC(LM_FTL_by_First_move5)
BIC(LM_FTL_by_First_move4)
BIC(LM_FTL_by_First_move3)
BIC(LM_FTL_by_First_move2)
BIC(LM_FTL_by_First_move1)
```
Le meilleur modèle est le plus simplifié
```{r}
LM_FTL_by_First_move <- LM_FTL_by_First_move8
```


### Treatment VS Teatment_bis

```{r}
AIC(lmer(log(Flight_to_light + 0.1) ~ First_move + Treatment_bis + Region +  Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) +  (1 | Room) + Region:Treatment_FTL, data = Data))
AIC(lmer(log(Flight_to_light + 0.1) ~ First_move + Treatment + Region +  Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) +  (1 | Room) + Region:Treatment_FTL, data = Data))
```

```{r}
summary(lmer(log(Flight_to_light + 0.1) ~ First_move + Treatment_bis + Region +  Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) +  (1 | Room) + Region:Treatment_FTL, data = Data))$coeff
summary(lmer(log(Flight_to_light + 0.1) ~ First_move + Treatment + Region +  Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) +  (1 | Room) + Region:Treatment_FTL, data = Data))$coeff
```
On garde Treatment_bis

### Résultats

Analyse des p-val
```{r}
summary(lmer(log(Flight_to_light + 0.1) ~ First_move + Treatment_bis + Region +  Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) +  (1 | Room) + Region:Treatment_FTL, data = Data))
summary(lmer(log(Flight_to_light + 0.1) ~ First_move + Treatment_bis + Region +  Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) +  (1 | Room) + Region:Treatment_FTL, data = Data_ref_smart))
summary(lmer(log(Flight_to_light + 0.1) ~ First_move + Treatment_bis + Region +  Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) +  (1 | Room) + Region:Treatment_FTL, data = Data_ref_full))
summary(lmer(log(Flight_to_light + 0.1) ~ First_move + Treatment_bis + Region +  Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) +  (1 | Room) + Region:Treatment_FTL, data = Data_ref_mixt_smart))
summary(lmer(log(Flight_to_light + 0.1) ~ First_move + Treatment_bis + Region +  Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) +  (1 | Room) + Region:Treatment_FTL, data = Data_refFTL_Mix))
summary(lmer(log(Flight_to_light + 0.1) ~ First_move + Treatment_bis + Region +  Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) +  (1 | Room) + Region:Treatment_FTL, data = Data_refFTL_Night))
```

```{r}
test1 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(Flight_to_light + 0.1) ~ First_move + Treatment_bis + Region +  Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) +  (1 | Room) + Region:Treatment_FTL, data = Data))$coeff)[,]))

test2 <- data.frame(Reference_Treatment = "Smart", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(Flight_to_light + 0.1) ~ First_move + Treatment_bis + Region +  Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) +  (1 | Room) + Region:Treatment_FTL, data = Data_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference_Treatment = "Full", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(Flight_to_light + 0.1) ~ First_move + Treatment_bis + Region +  Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) +  (1 | Room) + Region:Treatment_FTL, data = Data_ref_full))$coeff)[,]))

test4 <- data.frame(Reference_Treatment = "Mixt_smart", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(Flight_to_light + 0.1) ~ First_move + Treatment_bis + Region +  Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) +  (1 | Room) + Region:Treatment_FTL, data = Data_ref_mixt_smart))$coeff)[,]))

test5 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Mixt", as.data.frame(( summary(lmer(log(Flight_to_light + 0.1) ~ First_move + Treatment_bis + Region +  Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) +  (1 | Room) + Region:Treatment_FTL, data = Data_refFTL_Mix))$coeff)[,]))

test6 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Night", as.data.frame(( summary(lmer(log(Flight_to_light + 0.1) ~ First_move + Treatment_bis + Region +  Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) +  (1 | Room) + Region:Treatment_FTL, data = Data_refFTL_Night))$coeff)[,]))

Results_FTL_by_first_move <- rbind(test1,test2,test3,test4,test5,test6)

colnames(Results_FTL_by_first_move ) <- c("Reference_Treatment", "Reference_Timing" ,"Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_FTL_by_first_move , file = "output/Results_FTL_by_first_move.csv" )

kable(Results_FTL_by_first_move , caption = " Temps mis pour voler à la lumière en fonction du temps mis pour bouger ", padding = 1 )
```

```{r}
Info_Flight_to_light_by_First_move <- "
En moyenne, au plus un individu a bougé rapidement, au plus il volera à la lampe rapidement.
Les individus expérimentés de jours avec 30 min à l'obscurité mettent significativement moins de temps pour atteindre la lampe par rapport aux individus testés à d'autres timing.
Les individus expérimentés de jours avec 30 min à l'obscurité descendant d'une mère provenant du luxembourg  mettent significativement plus de temps pour atteindre la lampe par rapport aux mêmes individus descendants d'une mère provenant du Brabant.
Les individus mettent quasiement significativement (p-val = 0.0749) plus de temps pour aller à la lampe quand elle est au Sud.
Les individus descendant d'une mère provenant de la région luxembourgeoise et expérimentés la nuit mettent significativement moins de temps pour voler à la lumière.
Les individus du traitement Mixt-smart mettent significativement moins de temps pour voler à la lampe que les individus du traitement Smart.
"
```


### Visualisation

```{r}
ggplot(data = Data, aes(x = log(First_move), y = log(Flight_to_light+0.1))) +
  geom_smooth(method = "lm", aes(color = Treatment_FTL), se = T, alpha = 0.2) +
  geom_point(aes(color = Treatment_FTL, shape = Treatment_bis), alpha = 0.75) +
  labs(title = "Temps pour voler à la lumière en fonction du temps pour faire le premier mouvement", x = "log(Temps pour le premier mouvement (s))", y = "log(Temps pour voler à la lumière (s))", color = "Timing", shape = "Traitement") +
  theme_bw() +
  theme(title=element_text(size=10))
```



## STL_by_First_move

### Test


```{r}
LM_Speed_to_light_by_First_move <- lmer(log(Speed_to_light+0.000001) ~ First_move + Treatment_bis + Sex  + Region + Treatment_FTL + Light_orientation + (1|Mother) + (1|Site) + (1|Room) , data = Data)
summary(LM_Speed_to_light_by_First_move)
```

```{r}
qqnorm(resid(LM_Speed_to_light_by_First_move))
qqline(resid(LM_Speed_to_light_by_First_move))
```


Homogénité des variances
```{r}
plot(LM_Speed_to_light_by_First_move)
```


Distance de cook
```{r}
plot(cooks.distance(LM_Speed_to_light_by_First_move))
```

### Simplification du modèle

On est partit d'un modèle avec aucune interaction aussi non le modèle devient trop complexe pour très peu d'échantillons.

### Treatment VS Treatment_bis

```{r}
AIC(lmer(log(Speed_to_light+0.000001) ~ First_move + Treatment_bis + Sex  + Region + Treatment_FTL + Light_orientation + (1|Mother) + (1|Site) + (1|Room) , data = Data))
AIC(lmer(log(Speed_to_light+0.000001) ~ First_move + Treatment + Sex  + Region + Treatment_FTL + Light_orientation + (1|Mother) + (1|Site) + (1|Room) , data = Data))
```


```{r}
summary(lmer(log(Speed_to_light+0.000001) ~ First_move + Treatment_bis + Sex  + Region + Treatment_FTL + Light_orientation + (1|Mother) + (1|Site) + (1|Room) , data = Data))$coeff
summary(lmer(log(Speed_to_light+0.000001) ~ First_move + Treatment + Sex  + Region + Treatment_FTL + Light_orientation + (1|Mother) + (1|Site) + (1|Room) , data = Data))$coeff
```

Il serait légèrement préférable de continuer avec Treatment_bis

### Résultats

Analyse des p-val

```{r}
summary(lmer(log(Speed_to_light+0.000001) ~ First_move +Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room), data = Data))
summary(lmer(log(Speed_to_light+0.000001) ~ First_move +Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_ref_full))
summary(lmer(log(Speed_to_light+0.000001) ~ First_move +Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_ref_smart))
summary(lmer(log(Speed_to_light+0.000001) ~ First_move +Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_ref_mixt_smart))
summary(lmer(log(Speed_to_light+0.000001) ~ First_move +Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_refFTL_Night))
summary(lmer(log(Speed_to_light+0.000001) ~ First_move +Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_refFTL_Mix))
```

```{r}
test1 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(Speed_to_light+0.000001) ~ First_move +Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room), data = Data))$coeff)[,]))

test2 <- data.frame(Reference_Treatment = "Smart", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(Speed_to_light+0.000001) ~ First_move +Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference_Treatment = "Full", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(Speed_to_light+0.000001) ~ First_move +Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_ref_full))$coeff)[,]))

test4 <- data.frame(Reference_Treatment = "Mixt_smart", Reference_Timing = "Day", as.data.frame(( summary(lmer(log(Speed_to_light+0.000001) ~ First_move +Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_ref_mixt_smart))$coeff)[,]))

test5 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Mixt", as.data.frame(( summary(lmer(log(Speed_to_light+0.000001) ~ First_move +Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_refFTL_Mix))$coeff)[,]))

test6 <- data.frame(Reference_Treatment = "Dark", Reference_Timing = "Night", as.data.frame(( summary(lmer(log(Speed_to_light+0.000001) ~ First_move +Treatment_bis + Sex + Region + Treatment_FTL + Light_orientation + (1 | Mother) + (1 | Site) + (1 | Room) , data = Data_refFTL_Night))$coeff)[,]))

Results_SPL_by_first_move <- rbind(test1,test2,test3,test4,test5,test6)

colnames(Results_SPL_by_first_move ) <- c("Reference_Treatment", "Reference_Timing" ,"Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_SPL_by_first_move , file = "output/Results_SPL_by_first_move.csv" )

kable(Results_SPL_by_first_move , caption = "Vitesse pour voler à la lumière en fonction du temps mis pour bouger", padding = 1 )
```

```{r}
Info_Speed_to_light_by_First_move <- "
Aucun effet significatif de First move.
Les individus venants du traitement Full sont quasiement significativement (p-val = 0.07514) plus rapide que ceux du traitement Dark.
Les individus expérimentés durant la nuit ou en timing mixte sont significativement plus rapide que ceux de jour.
Les individus étants expérimentés dans les conditions mixtes sont plus lent que les individus expérimentés durant la nuit mais pas de manière significative (p-val = 0.124253).
"
```


### Visualisation


```{r}
ggplot(data = Data, aes(x = log(First_move), y = log(Speed_to_light+0.000001))) +
  geom_smooth(method = "lm", aes(color = Treatment_FTL), se = T, alpha = 0.2) + 
  geom_point(aes(color = Treatment_FTL, shape = Treatment_bis), alpha = 0.75) +
  labs(title = "Vitesse de déplacement à la lumière en fonction du temps pour le premier mouvement", x = "log(Temps mis pour faire le premier mouvement (s) ", y = "log(Vitesse pour aller à la lumière (min/s))", color = "Timing", shape = "Traitement") +
  theme_bw() +
  theme(title=element_text(size=10))
```




# Split by Species



## Total_molt OP

Vu que c'est une variable de comptage, nous comptons faire un glm de type poisson

### Test

```{r}
GLM_Total_molt_OP <- glmer(Total_molt ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region + Sex:Region , family = poisson, data = Data_OP)
summary(GLM_Total_molt_OP)
```
Pas d'effet significatifs

Vérification des hypothèses
```{r}
resid_panel(GLM_Total_molt_OP, plots = c("resid"), smoother = TRUE)
```

Le modèle pourrait être siplifié par AIC (surtout pour les interactions)

### Sipmplification du modèle

Simplification par AIC/BIC
```{r}
drop1(GLM_Total_molt_OP)
```
```{r}
GLM_Total_molt2_OP <- update(GLM_Total_molt_OP, . ~ . - Treatment_bis:Sex)
drop1(GLM_Total_molt2_OP)
```
```{r,cache=TRUE}
GLM_Total_molt3_OP <- update(GLM_Total_molt2_OP, . ~ . - Treatment_bis:Region)
drop1(GLM_Total_molt3_OP)
```

Modèle final, aucune interactions n'était significative

```{r}
GLM_Total_molt_OP <- GLM_Total_molt3_OP
summary(GLM_Total_molt_OP)
```


### Treatment VS Treatment_bis

```{r}
AIC(GLM_Total_molt_OP)
AIC(update(GLM_Total_molt_OP, . ~ . - Treatment_bis + Treatment))
```
```{r}
summary(GLM_Total_molt_OP)$coeff
summary(update(GLM_Total_molt_OP, . ~ . - Treatment_bis + Treatment))$coeff
```

Aucune différence de significativité entre treatment et treatment_bis, on peut garder treatment_bis qui est meilleur scientifiquement parlant.

### Résultats

```{r, message=FALSE}
summary(glmer(Total_molt ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Sex:Region, family = poisson, data = Data_OP))
summary(glmer(Total_molt ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Sex:Region, family = poisson, data = Data_OP_ref_smart))
summary(glmer(Total_molt ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Sex:Region, family = poisson, data = Data_OP_ref_full))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame((summary(glmer(Total_molt ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Sex:Region, family = poisson, data = Data_OP))$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame((summary(glmer(Total_molt ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Sex:Region, family = poisson, data = Data_OP_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame((summary(glmer(Total_molt ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Sex:Region, family = poisson, data = Data_OP_ref_full))$coeff)[,]))

Results_Total_molt_OP <- rbind(test1,test2,test3)

colnames(Results_Total_molt_OP) <- c("Reference","Estimate", "Std. Error", "z value", "Pr(>|z|)")

write.csv2(Results_Total_molt_OP, file = "output/Results_Total_molt_OP.csv")

kable(Results_Total_molt_OP, caption = "Nombre total de mues OP", padding = 1 )

```


```{r}
Info_Total_molt_OP <- "
Aucun effet significatif
"
Info_Total_molt_OP
```




## Total_molt AE

Vu que c'est une variable de comptage, nous comptons faire un glm de type poisson

### Test

```{r}
GLM_Total_molt_AE <- glmer(Total_molt ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region + Sex:Region , family = poisson, data = Data_AE)
summary(GLM_Total_molt_AE)
```
Pas d'effet significatifs

Vérification des hypothèses
```{r}
resid_panel(GLM_Total_molt_AE, plots = c("resid"), smoother = TRUE)
```

Le modèle pourrait être siplifié par AIC (surtout pour les interactions)

### Sipmplification du modèle

Simplification par AIC/BIC
```{r}
drop1(GLM_Total_molt_AE)
```
```{r}
GLM_Total_molt2_AE <- update(GLM_Total_molt_AE, . ~ . - Treatment_bis:Sex)
drop1(GLM_Total_molt2_AE)
```
```{r,cache=TRUE}
GLM_Total_molt3_AE <- update(GLM_Total_molt2_AE, . ~ . - Treatment_bis:Region)
drop1(GLM_Total_molt3_AE)
```

Modèle final, aucune interactions n'était significative

```{r}
GLM_Total_molt_AE <- GLM_Total_molt3_AE
summary(GLM_Total_molt_AE)
```


### Treatment VS Treatment_bis

```{r}
AIC(GLM_Total_molt_AE)
AIC(update(GLM_Total_molt_AE, . ~ . - Treatment_bis + Treatment))
```
```{r}
summary(GLM_Total_molt_AE)$coeff
summary(update(GLM_Total_molt_AE, . ~ . - Treatment_bis + Treatment))$coeff
```
Aucune différence de significativité entre treatment et treatment_bis

### Résultats
```{r}
summary(glmer(Total_molt ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Sex:Region, data = Data_AE, family = poisson))
summary(glmer(Total_molt ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Sex:Region, data = Data_AE_ref_smart, family = poisson))
summary(glmer(Total_molt ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Sex:Region, data = Data_AE_ref_full, family = poisson))
summary(glmer(Total_molt ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Sex:Region, data = Data_AE_ref_full, family = poisson))
```



```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(glmer(Total_molt ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Sex:Region, data = Data_AE, family = poisson))$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(glmer(Total_molt ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Sex:Region, data = Data_AE_ref_smart, family = poisson))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(glmer(Total_molt ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Sex:Region, data = Data_AE_ref_full, family = poisson))$coeff)[,]))

test4 <- data.frame(Reference = "Mixt_smart", as.data.frame(( summary(glmer(Total_molt ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Sex:Region, data = Data_AE_ref_mixt_smart, family = poisson))$coeff)[,])) 

Results_Total_molt_AE <- rbind(test1,test2,test3,test4)

colnames(Results_Total_molt_AE ) <- c("Reference","Estimate", "Std. Error", "z value", "Pr(>|z|)")

write.csv2(  Results_Total_molt_AE, file = "output/Results_Total_molt_AE.csv" )

kable(Results_Total_molt_AE , caption = " Nombre total de mues AE ", padding = 1 )
```

```{r}
Info_Total_molt_AE <- "
Aucun effet significatif
"
Info_Total_molt_AE
```




## Diapause OP

### Test

```{r, warning = F}
GLM_Diapause_OP <- glmer(Diapause ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_OP, family = binomial)
summary(GLM_Diapause_OP)
```

### Simplification du modèle

```{r, warning = F}
drop1(GLM_Diapause_OP)
```
```{r,warning=FALSE, message=FALSE}
GLM_Diapause1_OP <- update(GLM_Diapause_OP, . ~ . - Treatment_bis:Sex)
AIC(GLM_Diapause1_OP)
AIC(glmer(Diapause ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room), data = Data_OP, family = binomial))
```

```{r}
GLM_Diapause_OP <- glmer(Diapause ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room), data = Data_OP, family = binomial)
```

### Treatment VS Treatment_bis

```{r, warning = F}
AIC(glmer(Diapause ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room), data = Data_OP, family = binomial))
AIC(glmer(Diapause ~ Treatment + Sex + Region + (1|Mother) + (1|Site) + (1|Room), data = Data_OP, family = binomial))
```

```{r, warning = F}
summary(glmer(Diapause ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room), data = Data_OP, family = binomial))$coeff
summary(glmer(Diapause ~ Treatment + Sex + Region + (1|Mother) + (1|Site) + (1|Room), data = Data_OP, family = binomial))$coeff
```
Il vaut mieux garder treatment

### Résultats

Analyse des p-val

```{r, warning = F}
summary(glmer(Diapause ~ Treatment + Sex + Region + (1|Mother) + (1|Site) + (1|Room), data = Data_OP, family = binomial))
summary(glmer(Diapause ~ Treatment + Sex + Region + (1|Mother) + (1|Site) + (1|Room), data = Data_OP_ref_smart, family = binomial))
summary(glmer(Diapause ~ Treatment + Sex + Region + (1|Mother) + (1|Site) + (1|Room), data = Data_OP_ref_full, family = binomial))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(glmer(Diapause ~ Treatment + Sex + Region + (1|Mother) + (1|Site) + (1|Room), data = Data_OP, family = binomial))$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(glmer(Diapause ~ Treatment + Sex + Region + (1|Mother) + (1|Site) + (1|Room), data = Data_OP_ref_smart, family = binomial))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(glmer(Diapause ~ Treatment + Sex + Region + (1|Mother) + (1|Site) + (1|Room), data = Data_OP_ref_full, family = binomial))$coeff)[,]))

Results_Diapause_OP <- rbind(test1,test2,test3)

colnames(Results_Diapause_OP ) <- c("Reference","Estimate", "Std. Error", "z value", "Pr(>|z|)")

write.csv2( Results_Diapause_OP , file = "output/Results_Diapause_OP.csv" )

kable(Results_Diapause_OP , caption = " Diapause OP ", padding = 1 )
```

```{r}
Info_Diapause_OP <-"
Aucun des paramètres sélectionnés n'est significatif mais les individus du traitements Full ont quasiement significativements (p-val = 0.0539) une plus faible probabilité d'être en diapause que les individus du traitement Dark.
"
```



## Diapause AE

### Test

```{r, warning = F}
GLM_Diapause_AE <- glmer(Diapause ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_AE, family = binomial)
summary(GLM_Diapause_AE)
```

### Simplification du modèle

```{r, warning = F}
drop1(GLM_Diapause_AE)
```
```{r,warning=FALSE, message=FALSE}
GLM_Diapause1_AE <- update(GLM_Diapause_AE, . ~ . - Treatment_bis:Region)
drop1(GLM_Diapause1_AE)
```

```{r}
GLM_Diapause_AE <- glmer(Diapause ~ Treatment_bis + Sex + Region + Treatment_bis:Sex + (1|Mother) + (1|Site) + (1|Room), data = Data_AE, family = binomial)
```

### Treatment VS Treatment_bis

```{r, warning = F}
AIC(glmer(Diapause ~ Treatment_bis + Sex + Region + Treatment_bis:Sex + (1|Mother) + (1|Site) + (1|Room), data = Data_AE, family = binomial))
AIC(glmer(Diapause ~ Treatment + Sex + Region + Treatment:Sex + (1|Mother) + (1|Site) + (1|Room), data = Data_AE, family = binomial))
```

```{r, warning = F}
summary(glmer(Diapause ~ Treatment_bis + Sex + Region + Treatment_bis:Sex + (1|Mother) + (1|Site) + (1|Room), data = Data_AE, family = binomial))$coeff
summary(glmer(Diapause ~ Treatment + Sex + Region + Treatment:Sex + (1|Mother) + (1|Site) + (1|Room), data = Data_AE, family = binomial))$coeff
```
Il vaut mieux garder treatment

### Résultats

Analyse des p-val

```{r, warning = F}
summary(glmer(Diapause ~ Treatment + Sex + Region + Treatment_bis:Sex + (1|Mother) + (1|Site) + (1|Room), data = Data_AE, family = binomial))
summary(glmer(Diapause ~ Treatment + Sex + Region + Treatment_bis:Sex + (1|Mother) + (1|Site) + (1|Room), data = Data_AE_ref_full, family = binomial))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(glmer(Diapause ~ Treatment + Sex + Region + Treatment_bis:Sex + (1|Mother) + (1|Site) + (1|Room), data = Data_AE, family = binomial))$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(glmer(Diapause ~ Treatment + Sex + Region + Treatment_bis:Sex + (1|Mother) + (1|Site) + (1|Room), data = Data_AE_ref_full, family = binomial))$coeff)[,]))

Results_Diapause_AE <- rbind(test1,test2)

colnames(Results_Diapause_AE ) <- c("Reference","Estimate", "Std. Error", "z value", "Pr(>|z|)")

write.csv2( Results_Diapause_AE , file = "output/Results_Diapause_AE.csv" )

kable(Results_Diapause_AE , caption = " Diapause AE ", padding = 1 )
```

```{r}
Info_Diapause_AE <-"
Aucun des facteurs repris ne s'avère significatif.
"
```








## TT_pupation OP


### Test

```{r}
LM_TT_pupation_OP <- lmer(TT_pupation ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_OP)
summary(LM_TT_pupation_OP)
```

Normalité des résidus
```{r}
qqnorm(resid(LM_TT_pupation_OP))
qqline(resid(LM_TT_pupation_OP))
```


Homogénité des variances
```{r}
plot(LM_TT_pupation_OP)
```

Distance de cook
```{r}
plot(cooks.distance(LM_TT_pupation_OP))
```


### Simplification du modèle

```{r}
drop1(LM_TT_pupation_OP)
```
```{r}
LM_TT_pupation1_OP <- update(LM_TT_pupation_OP, . ~ . - Treatment_bis:Sex)
drop1(LM_TT_pupation1_OP)
```
```{r}
LM_TT_pupation2_OP <- update(LM_TT_pupation1_OP, . ~ . - Treatment_bis:Region)
drop1(LM_TT_pupation2_OP)
```
On a plus d'interactions, nous pouvons nous arrêter ici

```{r}
LM_TT_pupation_OP <- LM_TT_pupation2_OP
```




### Treatment VS Treatment_bis
```{r}
AIC(lmer(TT_pupation ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) , data = Data_OP))
AIC(lmer(TT_pupation ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) , data = Data_OP))
```

```{r}
summary(lmer(TT_pupation ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) , data = Data_OP))$coeff
summary(lmer(TT_pupation ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) , data = Data_OP))$coeff
```
Aucune différence significative

### Résultats

Analyse des P-val
```{r}
summary(lmer(TT_pupation ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) , data = Data_OP))
summary(lmer(TT_pupation ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) , data = Data_OP_ref_full))
summary(lmer(TT_pupation ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) , data = Data_OP_ref_smart))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(lmer(TT_pupation ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) , data = Data_OP))
$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(lmer(TT_pupation ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) , data = Data_OP_ref_smart))
$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(lmer(TT_pupation ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) , data = Data_OP_ref_full))
$coeff)[,]))

Results_TT_pupation_OP <- rbind(test1,test2,test3)

colnames(Results_TT_pupation_OP ) <- c("Reference","Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_TT_pupation_OP , file = "output/Results_TT_pupation_OP.csv" )

kable(Results_TT_pupation_OP , caption = " Temps pour entrer en chrysalide OP ", padding = 1 )
```

```{r}
Infos_TT_pupation_OP <- "
Mâle presque significativement plus rapide mais c'est tout
"
Infos_TT_pupation_OP
```



## TT_pupation AE


### Test

```{r}
LM_TT_pupation_AE <- lmer(TT_pupation ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_AE)
summary(LM_TT_pupation_AE)
```

Normalité des résidus
```{r}
qqnorm(resid(LM_TT_pupation_AE))
qqline(resid(LM_TT_pupation_AE))
```
on va changer les variables pour la normlité

```{r}
LM_TT_pupation_AE <- lmer(log(TT_pupation) ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_AE)
summary(LM_TT_pupation_AE)
```

Normalité des résidus
```{r}
qqnorm(resid(LM_TT_pupation_AE))
qqline(resid(LM_TT_pupation_AE))
```
On a vu mieux mais c'est passable

Homogénité des variances
```{r}
plot(LM_TT_pupation_AE)
```

Distance de cook
```{r}
plot(cooks.distance(LM_TT_pupation_AE))
```


### Simplification du modèle

```{r}
drop1(LM_TT_pupation_AE)
```
```{r}
LM_TT_pupation1_AE <- update(LM_TT_pupation_AE, . ~ . - Treatment_bis:Sex)
drop1(LM_TT_pupation1_AE)
```

On a plus d'interactions, nous pouvons nous arrêter ici

```{r}
LM_TT_pupation_AE <- LM_TT_pupation1_AE
```




### Treatment VS Treatment_bis
```{r}
AIC(lmer(log(TT_pupation) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE))
AIC(lmer(log(TT_pupation) ~ Treatment + Sex + Region + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment:Region, data = Data_AE))
```

```{r}
summary(lmer(log(TT_pupation) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE))$coeff
summary(lmer(log(TT_pupation) ~ Treatment + Sex + Region + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment:Region, data = Data_AE))$coeff
```
Treatment_bis est mieux mais il y a aucune différence significatives

### Résultats

Analyse des P-val
```{r}
summary(lmer(log(TT_pupation) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE))
summary(lmer(log(TT_pupation) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE_ref_full))
summary(lmer(log(TT_pupation) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE_ref_smart))
summary(lmer(log(TT_pupation) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE_ref_mixt_smart))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(lmer(log(TT_pupation) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE))$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(lmer(log(TT_pupation) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(lmer(log(TT_pupation) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE_ref_full))$coeff)[,]))

test4 <- data.frame(Reference = "Mixt_smart", as.data.frame(( summary(lmer(log(TT_pupation) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE_ref_mixt_smart))$coeff)[,]))

Results_TT_puapation_AE <- rbind(test1,test2,test3, test4)

colnames(Results_TT_puapation_AE ) <- c("Reference","Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_TT_puapation_AE , file = "output/Results_TT_puapation_AE.csv" )

kable(Results_TT_puapation_AE , caption = " Temps pour entrer en chrysalide AE ", padding = 1 )
```

```{r}
Infos_TT_pupation_AE <- "

Le traitement Mixt-smart en région luxembourgoise est significativement plus lent que l'effet de la région luxembourgeoise et du traitement mixt-smart additionné sauf quand on le compare au traitement full où là, la région luxembourg est elle même significativement plus lent
Le traitement Smart en région luxembourgeoise est significativement plus rapide que l'effet de la région luxembourgeoise et du traitement smart additioné quand il est comparé au traitement full.
"
Infos_TT_pupation_AE
```








## Time_in_pupa OP

### Test

```{r}
LM_Time_in_pupa_OP <- lmer(Time_in_pupa ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_OP)
summary(LM_TT_pupation_OP)
```

Vérification de la normalité
```{r}
qqnorm(resid(LM_Time_in_pupa_OP))
qqline(resid(LM_Time_in_pupa_OP))
```
Je pense que cela est suffisemment bon

Homogénité des variances
```{r}
plot(LM_Time_in_pupa_OP)
```

Distance de cook
```{r}
plot(cooks.distance(LM_Time_in_pupa_OP))
```

### Simplification du modèle

```{r}
drop1(LM_Time_in_pupa_OP)
```
```{r}
LM_Time_in_pupa1_OP <- update(LM_Time_in_pupa_OP, . ~ . - Treatment_bis:Region)
drop1(LM_Time_in_pupa1_OP)
```
```{r}
LM_Time_in_pupa2_OP <- update(LM_Time_in_pupa1_OP, . ~ . - Treatment_bis:Sex)
drop1(LM_Time_in_pupa2_OP)
```
Nous n'avons plus d'interactions, nous nous arrêtons là.

Regardons si les simplifications étaient utiles
```{r}
AIC(LM_Time_in_pupa2_OP)
AIC(LM_Time_in_pupa_OP)
```
Les simplifications étaient utiles
```{r}
LM_Time_in_pupa_OP <- LM_Time_in_pupa2_OP
```


### Treatment VS Treatment_bis

```{r}
AIC(lmer(Time_in_pupa ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room), data = Data_OP))
AIC(lmer(Time_in_pupa ~ Treatment + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room), data = Data_OP))
```
```{r}
summary(lmer(Time_in_pupa ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room), data = Data_OP))$coeff
summary(lmer(Time_in_pupa ~ Treatment + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room), data = Data_OP))$coeff
```

Selon les summary le choix de treatment ou treatment_bis est discutable mais on peut alors se pencher sur l'AIC qui nous encourage à prendre treatment_bis.

### Résultats

Analysons les P-val
```{r}
summary(LM_Time_in_pupa_OP)
summary(lmer(Time_in_pupa ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room), data = Data_OP_ref_full))
summary(lmer(Time_in_pupa ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) , data = Data_OP_ref_smart))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(LM_Time_in_pupa_OP)$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(lmer(Time_in_pupa ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) , data = Data_OP_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(lmer(Time_in_pupa ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room), data = Data_OP_ref_full))$coeff)[,]))

Results_Time_in_pupa_OP <- rbind(test1,test2,test3)

colnames(Results_Time_in_pupa_OP ) <- c("Reference","Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_Time_in_pupa_OP , file = "output/Results_Time_in_pupa_OP.csv" )

kable(Results_Time_in_pupa_OP , caption = " Temps en chrysalide OP ", padding = 1 )
```

```{r}
Info_time_in_pupa_OP <- "
Les mâmes passent significativement (0.0144) plus de temps (0.44 jours) en chrysaldie
Les individus du traitement Smart passent significativement (0.0280) moins de temps (0.61 jours) en chrysalide que les individus du traitement Full.
"
Info_time_in_pupa_OP
```





## Time_in_pupa AE

### Test

```{r}
LM_Time_in_pupa_AE <- lmer(Time_in_pupa ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_AE)
summary(LM_TT_pupation_AE)
```

Vérification de la normalité
```{r}
qqnorm(resid(LM_Time_in_pupa_AE))
qqline(resid(LM_Time_in_pupa_AE))
```
Ca a vraiment l'air nickel

Homogénité des variances
```{r}
plot(LM_Time_in_pupa_AE)
```

Distance de cook
```{r}
plot(cooks.distance(LM_Time_in_pupa_AE))
```

### Simplification du modèle

```{r}
drop1(LM_Time_in_pupa_AE)
```
```{r}
LM_Time_in_pupa1_AE <- update(LM_Time_in_pupa_AE, . ~ . - Treatment_bis:Sex)
drop1(LM_Time_in_pupa1_AE)
```
Le reste est significatif, nous devrions nous arrêter là

Regardons si les simplifications étaient utiles
```{r}
AIC(LM_Time_in_pupa1_AE)
AIC(LM_Time_in_pupa_AE)
```
Les simplifications étaient utiles
```{r}
LM_Time_in_pupa_AE <- LM_Time_in_pupa1_AE
```


### Treatment VS Treatment_bis

```{r}
AIC(lmer(Time_in_pupa ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE))
AIC(lmer(Time_in_pupa ~ Treatment + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment:Region, data = Data_AE))
```
```{r}
summary(lmer(Time_in_pupa ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE))$coeff
summary(lmer(Time_in_pupa ~ Treatment + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment:Region, data = Data_AE))$coeff
```

Il faut garder treatment_bis

### Résultats



Analysons les P-val
```{r}
summary(LM_Time_in_pupa_AE)
summary(lmer(Time_in_pupa ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE_ref_full))
summary(lmer(Time_in_pupa ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE_ref_smart))
summary(lmer(Time_in_pupa ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE_ref_mixt_smart))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(LM_Time_in_pupa_AE)$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(lmer(Time_in_pupa ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(lmer(Time_in_pupa ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE_ref_full))$coeff)[,]))

test4 <- data.frame(Reference = "Mixt_smart", as.data.frame(( summary(lmer(Time_in_pupa ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE_ref_mixt_smart))$coeff)[,]))

Results_Time_in_pupa_AE <- rbind(test1,test2,test3,test4)

colnames(Results_Time_in_pupa_AE ) <- c("Reference","Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_Time_in_pupa_AE , file = "output/Results_Time_in_pupa_AE.csv" )

kable(Results_Time_in_pupa_AE , caption = " Temps en chrysalide AE ", padding = 1 )
```

```{r}
Info_time_in_pupa_AE <- "
les individus du traitement Full passent significativement (p-val = 0.008934) plus de temps (0.6459 jours) en chrysalides que les individus du traitement Dark.
Les mâles passent significativement (p-val = 0.000951) plus de temps (0.51 jours) en chrysalides.
Les individus du traitement Mixt-smart en région luxembourgoise passe significativement plus de temps en chrysalide que l'effet de la région luxembourgeoise et du traitement mixt-smart additionnés.
"
Info_time_in_pupa_AE
```





## TT_emerge OP

### Test

```{r}
LM_TT_emerge_OP <- lmer(TT_emerge ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_OP)
summary(LM_TT_pupation_OP)
```


Vérification de la normalité
```{r}
qqnorm(resid(LM_TT_emerge_OP))
qqline(resid(LM_TT_emerge_OP))
```



Homogénité des variances
```{r}
plot(LM_TT_emerge_OP)
```

Distance de cook
```{r}
plot(cooks.distance(LM_TT_emerge_OP))
```

### Simplification du modèle


```{r}
drop1(LM_TT_emerge_OP)
```
```{r}
LM_TT_emerge1_OP <- update(LM_TT_emerge_OP, . ~ . - Treatment_bis:Sex)
drop1(LM_TT_emerge1_OP)
```
```{r}
LM_TT_emerge2_OP <- update(LM_TT_emerge1_OP, . ~ . - Treatment_bis:Region)
drop1(LM_TT_emerge2_OP)
```
Nous n'avons plus d'interactions, nous pouvons nous arrêter ici

Regardons si les simplifications étaient utiles
```{r}
AIC(LM_TT_emerge_OP)
AIC(LM_TT_emerge1_OP)
AIC(LM_TT_emerge2_OP)
```
Les simplifications n'ont pas étées utiles




### Treatment VS Treatment_bis
```{r}
AIC(lmer(TT_emerge ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_OP))
AIC(lmer(TT_emerge ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment:Sex + Treatment:Region, data = Data_OP))
```
```{r}
summary(lmer(TT_emerge ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_OP))$coeff
summary(lmer(TT_emerge ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment:Sex + Treatment:Region, data = Data_OP))$coeff
```
Treatment serait à privilégier au vu des AIC

### Résultats

Analysons les P-val

```{r}
summary(LM_TT_emerge_OP)
summary(lmer(TT_emerge ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment:Sex + Treatment:Region, data = Data_OP_ref_full))
summary(lmer(TT_emerge ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment:Sex + Treatment:Region, data = Data_OP_ref_smart))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(LM_TT_emerge_OP)$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(lmer(TT_emerge ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment:Sex + Treatment:Region, data = Data_OP_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(lmer(TT_emerge ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment:Sex + Treatment:Region, data = Data_OP_ref_full))$coeff)[,]))

Results_TT_emerge_OP <- rbind(test1,test2,test3)

colnames(Results_TT_emerge_OP ) <- c("Reference","Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_TT_emerge_OP , file = "output/Results_TT_emerge_OP.csv" )

kable(Results_TT_emerge_OP, caption = " Temps pour sortir de chrysalide OP ", padding = 1 )
```

```{r}
Infos_TT_emerge_OP <- "
RIEN DU TOUT de significatif
"
Infos_TT_emerge_OP
```



## TT_emerge AE

### Test

```{r}
LM_TT_emerge_AE <- lmer(TT_emerge ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_AE)
summary(LM_TT_pupation_AE)
```


Vérification de la normalité
```{r}
qqnorm(resid(LM_TT_emerge_AE))
qqline(resid(LM_TT_emerge_AE))
```
Cela pourrait encore aller peut-être mais on peut tenter en faisant des modifications

```{r}
LM_TT_emerge_AE <- lmer(log(TT_emerge) ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_AE)
```


Vérification de la normalité
```{r}
qqnorm(resid(LM_TT_emerge_AE))
qqline(resid(LM_TT_emerge_AE))
```
C'est déjà beaucoup mieux

Homogénité des variances
```{r}
plot(LM_TT_emerge_AE)
```

Distance de cook
```{r}
plot(cooks.distance(LM_TT_emerge_AE))
```

Je pense que je vais retirer le point outlier

```{r}
Data_TT_emerge_AE <- (Data_AE %>% filter(TT_emerge != "NA"))[unname(cooks.distance(LM_TT_emerge_AE) != max(cooks.distance(LM_TT_emerge_AE))),]

Data_TT_emerge_AE_ref_smart <- Data_TT_emerge_AE
Data_TT_emerge_AE_ref_smart$Treatment_bis <- relevel(Data_TT_emerge_AE$Treatment_bis, ref = "Smart")

Data_TT_emerge_AE_ref_full <- Data_TT_emerge_AE
Data_TT_emerge_AE_ref_full$Treatment_bis <- relevel(Data_TT_emerge_AE$Treatment_bis, ref = "Full")

Data_TT_emerge_AE_ref_mixt_smart <- Data_TT_emerge_AE
Data_TT_emerge_AE_ref_mixt_smart$Treatment_bis <- relevel(Data_TT_emerge_AE$Treatment_bis, ref = "Mixt_smart")
```

```{r}
LM_TT_emerge_AE <- lmer(log(TT_emerge) ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_TT_emerge_AE)
```


Vérification de la normalité
```{r}
qqnorm(resid(LM_TT_emerge_AE))
qqline(resid(LM_TT_emerge_AE))
```

Homogénité des variances
```{r}
plot(LM_TT_emerge_AE)
```

Distance de cook
```{r}
plot(cooks.distance(LM_TT_emerge_AE))
```


### Simplification du modèle


```{r}
drop1(LM_TT_emerge_AE)
```
```{r}
LM_TT_emerge1_AE <- update(LM_TT_emerge_AE, . ~ . - Treatment_bis:Sex)
drop1(LM_TT_emerge1_AE)
```

Nous n'avons plus d'interactions non significatives, nous pouvons nous arrêter ici

Regardons si les simplifications étaient utiles
```{r}
AIC(LM_TT_emerge_AE)
AIC(LM_TT_emerge1_AE)
```
Les simplifications ont étées utiles


### Treatment VS Treatment_bis
```{r}
AIC(lmer(log(TT_emerge) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_TT_emerge_AE))
AIC(lmer(log(TT_emerge) ~ Treatment + Sex + Region + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment:Region, data = Data_TT_emerge_AE))
```
```{r}
summary(lmer(log(TT_emerge) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_TT_emerge_AE))$coeff
summary(lmer(log(TT_emerge) ~ Treatment + Sex + Region + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment:Region, data = Data_TT_emerge_AE))$coeff
```
Treatment_bis serait à privilégier au vu des AIC

### Résultats

Analysons les P-val

```{r}
summary(lmer(log(TT_emerge) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_TT_emerge_AE))
summary(lmer(log(TT_emerge) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_TT_emerge_AE_ref_smart))
summary(lmer(log(TT_emerge) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_TT_emerge_AE_ref_full))
summary(lmer(log(TT_emerge) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_TT_emerge_AE_ref_mixt_smart))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(lmer(log(TT_emerge) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_TT_emerge_AE))$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(lmer(log(TT_emerge) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_TT_emerge_AE_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(lmer(log(TT_emerge) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_TT_emerge_AE_ref_full))$coeff)[,]))

test4 <- data.frame(Reference = "Mixt_smart", as.data.frame(( summary(lmer(log(TT_emerge) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_TT_emerge_AE_ref_mixt_smart))$coeff)[,]))

Results_TT_emerge_AE <- rbind(test1,test2,test3,test4)

colnames(Results_TT_emerge_AE ) <- c("Reference","Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_TT_emerge_AE , file = "output/Results_TT_emerge_AE.csv" )

kable(Results_TT_emerge_AE , caption = " Temps pour sortir de chrysalide AE ", padding = 1 )
```

```{r}
Infos_TT_emerge_AE <- "
Le traitement mixt-smart en région luxembourgoise est significativement plus lent que l'effet de la région luxembourgeoise et du traitement mixt-smart additionné sauf quand on le compare au traitement full où là, la région luxembourg est elle même significativement plus lente.
Les individus du traitement full mettent significativements (p-val = 0.0245) plus de temps (1.17 jours) à atteindre le stade adulte si ils descendent d'une mère qui provient du luxembourg.
Le traitement full en région luxembourgoise est significativement plus lent que l'effet de la région luxembourgeoise et du traitement full additionné qaund ils sont comparés au traitement smart.
"
Infos_TT_emerge_AE
```



## Pupa_mass OP

### Test

```{r}
LM_Pupa_mass_OP <- lmer(Pupa_mass ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_OP)
summary(LM_Pupa_mass_OP)
```

Vérification des hypothèses

```{r}
qqnorm(resid(LM_Pupa_mass_OP))
qqline(resid(LM_Pupa_mass_OP))
```


Homogénité des variances
```{r}
plot(LM_Pupa_mass_OP)
```

Distance de cook
```{r}
plot(cooks.distance(LM_Pupa_mass_OP))
```

### Simplification du modèle

```{r}
drop1(LM_Pupa_mass_OP)
```
```{r}
LM_Pupa_mass1_OP <- update(LM_Pupa_mass_OP, . ~ . - Treatment_bis:Region)
drop1(LM_Pupa_mass1_OP)
```

Il n'y a plus aucune interaction non significative, nous arrêtons la simplification du modèle

Regardons si les simplifications étaient utiles
```{r}
AIC(LM_Pupa_mass1_OP)
AIC(LM_Pupa_mass_OP)
```
Elle l'a été

```{r}
LM_Pupa_mass <- LM_Pupa_mass1_OP
```

### Treatment VS Treatment_bis

```{r}
AIC(lmer(Pupa_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex, data = Data_OP))
AIC(lmer(Pupa_mass ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment:Sex, data = Data_OP))
```
```{r}
summary(lmer(Pupa_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex, data = Data_OP))$coeff
summary(lmer(Pupa_mass ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment:Sex, data = Data_OP))$coeff
```
Treatment à l'air d'être meilleur

### Résultats

Analyse des p-val

```{r}
summary(lmer(Pupa_mass ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment:Sex, data = Data_OP))
summary(lmer(Pupa_mass ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment:Sex, data = Data_OP_ref_smart))
summary(lmer(Pupa_mass ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment:Sex, data = Data_OP_ref_full))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(lmer(Pupa_mass ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment:Sex, data = Data_OP))$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(lmer(Pupa_mass ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment:Sex, data = Data_OP_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(lmer(Pupa_mass ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment:Sex, data = Data_OP_ref_full))$coeff)[,]))

Results_Pupa_mass_OP <- rbind(test1,test2,test3)

colnames(Results_Pupa_mass_OP ) <- c("Reference","Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_Pupa_mass_OP , file = "output/Results_Pupa_mass_OP.csv" )

kable(Results_Pupa_mass_OP , caption = " Poids des chrysalides OP ", padding = 1 )
```

```{r}
Info_Pupa_mass_OP <- "
Les chrysalides du traitement Full sont quasiment significativement (p-val = 0.0680) plus légères (0.0085 g)  que celels du traitement Dark.
Les chrysalides des individus mâles du traitement Smart sont significativement (p-val = 0.0223) plus lourdes (0.015 g) que l'effet de mâle et du traitement Smart additioné comparé au traitement Full.
Les mâles sont significativements (p-val = 0.041) plus lourd (0.010 g) que les femelles dans le traitement Smart.
"
Info_Pupa_mass_OP
```




## Pupa_mass AE

### Test

```{r}
LM_Pupa_mass_AE <- lmer(Pupa_mass ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_AE)
summary(LM_Pupa_mass_AE)
```

Vérification des hypothèses

```{r}
qqnorm(resid(LM_Pupa_mass_AE))
qqline(resid(LM_Pupa_mass_AE))
```


Homogénité des variances
```{r}
plot(LM_Pupa_mass_AE)
```

Distance de cook
```{r}
plot(cooks.distance(LM_Pupa_mass_AE))
```

### Simplification du modèle

```{r}
drop1(LM_Pupa_mass_AE)
```
```{r}
LM_Pupa_mass1_AE <- update(LM_Pupa_mass_AE, . ~ . - Treatment_bis:Region)
drop1(LM_Pupa_mass1_AE)
```

Il n'y a plus aucune interaction non significative, nous arrêtons la simplification du modèle

Regardons si les simplifications étaient utiles
```{r}
AIC(LM_Pupa_mass1_AE)
AIC(LM_Pupa_mass_AE)
```
Elle l'a été

```{r}
LM_Pupa_mass <- LM_Pupa_mass1_AE
```

### Treatment VS Treatment_bis

```{r}
AIC(lmer(Pupa_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex, data = Data_AE))
AIC(lmer(Pupa_mass ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment:Sex, data = Data_AE))
```
```{r}
summary(lmer(Pupa_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex, data = Data_AE))$coeff
summary(lmer(Pupa_mass ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment:Sex, data = Data_AE))$coeff
```
Treatment à l'air d'être meilleur pour expliquer les données (AIc) mais le traitement Mixt Smart est fort significatif donc je préfère garder Treatment_bis

### Résultats

Analyse des p-val

```{r}
summary(lmer(Pupa_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex, data = Data_AE))
summary(lmer(Pupa_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex, data = Data_AE_ref_smart))
summary(lmer(Pupa_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex, data = Data_AE_ref_full))
summary(lmer(Pupa_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex, data = Data_AE_ref_mixt_smart))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(lmer(Pupa_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex, data = Data_AE))$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(lmer(Pupa_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex, data = Data_AE_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(lmer(Pupa_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex, data = Data_AE_ref_full))$coeff)[,]))

test4 <- data.frame(Reference = "Mixt_smart", as.data.frame(( summary(lmer(Pupa_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex, data = Data_AE_ref_mixt_smart))$coeff)[,]))

Results_Pupa_mass_AE <- rbind(test1,test2,test3,test4)

colnames(Results_Pupa_mass_AE ) <- c("Reference","Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_Pupa_mass_AE , file = "output/Results_Pupa_mass_AE.csv" )

kable(Results_Pupa_mass_AE , caption = " Poids des chrysalides AE ", padding = 1 )
```

```{r}
Info_Pupa_mass_AE <- "
Les individus du traitement Mixt-smart sont significfativement (p-val = 5.97e-05) plus lourds (0.078 g) que ceux du traitement Dark mais les mâles (toujours comparé au traitement Dark) le sont significativvement (p-val = 0.0297) moins  que les femelles (0.063 g).
Les chrysalides des individus du traitement Mixt smart sont significativements (p-val = 0.00259) plus lourd (0.061 g) que celles du traitement Smart.
Les chrysalides du traitement Mixt-smart sont significfativements (p-val = 0.00248) plus lourdes (0.058 g) que celles du traitement Full mais les mâles le sont significativvement (p-val = 0.02809) moins (0.062 g) que les femelles.
"
Info_Pupa_mass_AE
```



## Adult_mass OP

### Test

```{r}
LM_Adult_mass_OP <- lmer(Adult_mass ~ Treatment_bis + Sex  + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_OP)
summary(LM_Adult_mass_OP)
```

Vérification des hypothèses

```{r}
qqnorm(resid(LM_Adult_mass_OP))
qqline(resid(LM_Adult_mass_OP))
```




Homogénité des variances
```{r}
plot(LM_Adult_mass_OP)
```

Distance de cook
```{r}
plot(cooks.distance(LM_Adult_mass_OP))
```


### Simplification du modèle

```{r}
drop1(LM_Adult_mass_OP)
```
```{r}
LM_Adult_mass1_OP <- update(LM_Adult_mass_OP, . ~ . - Treatment_bis:Region)
drop1(LM_Adult_mass1_OP)
```
```{r}
LM_Adult_mass2_OP <- update(LM_Adult_mass1_OP, . ~ . - Treatment_bis:Sex)
drop1(LM_Adult_mass2_OP)
```

Nous n'avons plus d'interaction, nous arrêtons la simplification du modèle.

Est ce que la simplification a été utile ?
```{r}
AIC(LM_Adult_mass2_OP)
AIC(LM_Adult_mass_OP)
```
Oui

```{r}
LM_Adult_mass_OP <- LM_Adult_mass2_OP
```


### Treatment VS Treatment_bis

```{r}
AIC(lmer(Adult_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_OP))
AIC(lmer(Adult_mass ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_OP))
```
```{r}
summary(lmer(Adult_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_OP))$coeff
summary(lmer(Adult_mass ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_OP))$coeff
```

On va continuer avec treatment_bis

### Résultats

Analyse des P-val
```{r}
summary(lmer(Adult_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_OP))
summary(lmer(Adult_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_OP_ref_smart))
summary(lmer(Adult_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_OP_ref_full))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(lmer(Adult_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_OP))$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(lmer(Adult_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_OP_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(lmer(Adult_mass ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_OP_ref_full))$coeff)[,]))

Results_Adult_mass_OP <- rbind(test1,test2,test3)

colnames(Results_Adult_mass_OP ) <- c("Reference","Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_Adult_mass_OP , file = "output/Results_Adult_mass_OP.csv" )

kable(Results_Adult_mass_OP , caption = " Poids des adultes OP ", padding = 1 )
```

```{r}
Info_Adult_mass_OP <- "
Les adultes du traitement Full sont significativement plus lourd que ceux du traitement Dark.
Les mâles sont significativement plus légers que les femelles
"
Info_Adult_mass_OP
```




## Adult_mass AE

### Test

```{r}
LM_Adult_mass_AE <- lmer(Adult_mass ~ Treatment_bis + Sex  + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_AE)
summary(LM_Adult_mass_AE)
```

Vérification des hypothèses

```{r}
qqnorm(resid(LM_Adult_mass_AE))
qqline(resid(LM_Adult_mass_AE))
```

On va insérer un logarythme dans les données quand même
```{r}
LM_Adult_mass_AE <- lmer(log(Adult_mass) ~ Treatment_bis + Sex  + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_AE)
```

Vérification des hypothèses

```{r}
qqnorm(resid(LM_Adult_mass_AE))
qqline(resid(LM_Adult_mass_AE))
```

Homogénité des variances
```{r}
plot(LM_Adult_mass_AE)
```

Distance de cook
```{r}
plot(cooks.distance(LM_Adult_mass_AE))
```


### Simplification du modèle

```{r}
drop1(LM_Adult_mass_AE)
```
```{r}
LM_Adult_mass1_AE <- update(LM_Adult_mass_AE, . ~ . - Treatment_bis:Region)
drop1(LM_Adult_mass1_AE)
```
Nous n'avons plus d'interaction à retirer, nous arrêtons la simplification du modèle.

Est ce que la simplification a été utile ?
```{r}
AIC(LM_Adult_mass1_AE)
AIC(LM_Adult_mass_AE)
```
Oui

```{r}
LM_Adult_mass_AE <- LM_Adult_mass1_AE
```


### Treatment VS Treatment_bis

```{r}
AIC(lmer(log(Adult_mass) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_AE))
AIC(lmer(log(Adult_mass) ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_AE))
```
```{r}
summary(lmer(log(Adult_mass) ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_AE))$coeff
summary(lmer(log(Adult_mass) ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_AE))$coeff
```
On va continuer avec treatment_bis car même si l'AIC préfère le modèle Treatment , Mixt smart a un effet significatif avec Treatment_bis

### Résultats

Analyse des P-val
```{r}
summary(lmer(log(Adult_mass) ~ Treatment_bis + Sex + Region + Treatment_bis:Sex + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_AE))
summary(lmer(log(Adult_mass) ~ Treatment_bis + Sex + Region + Treatment_bis:Sex + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_AE_ref_smart))
summary(lmer(log(Adult_mass) ~ Treatment_bis + Sex + Region + Treatment_bis:Sex + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_AE_ref_full))
summary(lmer(log(Adult_mass) ~ Treatment_bis + Sex + Region + Treatment_bis:Sex + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_AE_ref_mixt_smart))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(lmer(log(Adult_mass) ~ Treatment_bis + Sex + Region + Treatment_bis:Sex + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_AE))$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(lmer(log(Adult_mass) ~ Treatment_bis + Sex + Region + Treatment_bis:Sex + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_AE_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(lmer(log(Adult_mass) ~ Treatment_bis + Sex + Region + Treatment_bis:Sex + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_AE_ref_full))$coeff)[,]))

test4 <- data.frame(Reference = "Mixt_smart", as.data.frame(( summary(lmer(log(Adult_mass) ~ Treatment_bis + Sex + Region + Treatment_bis:Sex + (1 | Mother) + (1 |  Site) + (1 | Room), data = Data_AE_ref_mixt_smart))$coeff)[,]))

Results_Adult_mass_AE <- rbind(test1,test2,test3,test4)

colnames(Results_Adult_mass_AE ) <- c("Reference","Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_Adult_mass_AE , file = "output/Results_Adult_mass_AE.csv" )

kable(Results_Adult_mass_AE , caption = " Poids des adultes AE ", padding = 1 )
```

```{r}
Info_Adult_mass_AE <- "
Les individus du traitement Full sont significativement plus lourds que ceux du traitement Dark.
Les individus du traitement Mixt-smart sont significativement plus lourd que ceux de n'importe quel autre traitement.
Les mâles significativement plus léger que les femelles dans le traitement Smart.
"
Info_Adult_mass_AE
```


## M0 OP

### Test

```{r}
LM_M0_OP <- lmer(M0 ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_OP)
summary(LM_M0_OP)
```

Vérification des hypothèses

```{r}
qqnorm(resid(LM_M0_OP))
qqline(resid(LM_M0_OP))
```
Je pense que ca va


Homogénité des variances
```{r}
plot(LM_M0_OP)
```


Distance de cook
```{r}
plot(cooks.distance(LM_M0_OP))
```


### Simplification du modèle

```{r}
drop1(LM_M0_OP)
```
```{r}
LM_M0_1_OP <- update(LM_M0_OP, . ~ . - Treatment_bis:Region)
drop1(LM_M0_1_OP)
```
### Résultats

```{r}
summary(LM_M0_1_OP)
summary(lmer(M0 ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) +  (1 | Room) + Treatment_bis:Sex, data = Data_OP_ref_smart))
summary(lmer(M0 ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) +  (1 | Room) + Treatment_bis:Sex, data = Data_OP_ref_full))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(LM_M0_1_OP)$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(lmer(M0 ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) +  (1 | Room) + Treatment_bis:Sex, data = Data_OP_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(lmer(M0 ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) +  (1 | Room) + Treatment_bis:Sex, data = Data_OP_ref_full))$coeff)[,]))

Results_M0_OP <- rbind(test1,test2,test3)

colnames(Results_M0_OP ) <- c("Reference","Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_M0_OP , file = "output/Results_M0_OP.csv" )

kable(Results_M0_OP , caption = " Poids initial OP ", padding = 1 )
```
Aucune différence significative de poids initial en fonction des traitements ou région ou sexe



## M0 AE

### Test

```{r}
LM_M0_AE <- lmer(M0 ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_AE)
summary(LM_M0_AE)
```

Vérification des hypothèses

```{r}
qqnorm(resid(LM_M0_AE))
qqline(resid(LM_M0_AE))
```
Je pense que ca va


Homogénité des variances
```{r}
plot(LM_M0_AE)
```


Distance de cook
```{r}
plot(cooks.distance(LM_M0_AE))
```


### Simplification du modèle

```{r}
drop1(LM_M0_AE)
```
```{r}
LM_M0_1_AE <- update(LM_M0_AE, . ~ . - Treatment_bis:Region)
drop1(LM_M0_1_AE)
```
### Résultats

```{r}
summary(LM_M0_1_AE)
summary(lmer(M0 ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) +  (1 | Room) + Treatment_bis:Sex, data = Data_AE_ref_smart))
summary(lmer(M0 ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) +  (1 | Room) + Treatment_bis:Sex, data = Data_AE_ref_full))
summary(lmer(M0 ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) +  (1 | Room) + Treatment_bis:Sex, data = Data_AE_ref_mixt_smart))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(LM_M0_1_AE)$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(lmer(M0 ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) +  (1 | Room) + Treatment_bis:Sex, data = Data_AE_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(lmer(M0 ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) +  (1 | Room) + Treatment_bis:Sex, data = Data_AE_ref_full))$coeff)[,]))

test4 <- data.frame(Reference = "Mixt_smart", as.data.frame(( summary(lmer(M0 ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 | Site) +  (1 | Room) + Treatment_bis:Sex, data = Data_AE_ref_mixt_smart))$coeff)[,]))

Results_M0_AE <- rbind(test1,test2,test3,test4)

colnames(Results_M0_AE ) <- c("Reference","Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_M0_AE , file = "output/Results_M0_AE.csv" )

kable(Results_M0_AE , caption = " Poids inital AE ", padding = 1 )
```
Aucune différence significative de poids initial en fonction des traitements ou région ou sexe





## Total_eaten OP

### Test

```{r}
LM_Total_eaten_OP <- lmer(Total_eaten ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_OP)
summary(LM_Total_eaten_OP)
```

Vérification des hypothèses
```{r}
qqnorm(resid(LM_Total_eaten_OP))
qqline(resid(LM_Total_eaten_OP))
```
C'est très bon.

Homogénité des variances
```{r}
plot(LM_Total_eaten_OP)
```
c'est correcte

Distance de cook
```{r}
plot(cooks.distance(LM_Total_eaten_OP))
```
Un point semble énormément influencer le modèle, je pense qu'il est préférable de le retirer

```{r}
Data_Total_eaten_OP <- (Data_OP %>% filter(Sex != "NA"))[unname(cooks.distance(LM_Total_eaten_OP) != max(cooks.distance(LM_Total_eaten_OP))),]

Data_Total_eaten_OP_ref_smart <- Data_Total_eaten_OP
Data_Total_eaten_OP_ref_smart$Treatment_bis <- relevel(Data_Total_eaten_OP$Treatment_bis, ref = "Smart")

Data_Total_eaten_OP_ref_full <- Data_Total_eaten_OP
Data_Total_eaten_OP_ref_full$Treatment_bis <- relevel(Data_Total_eaten_OP$Treatment_bis, ref = "Full")
```

et on refait les test

```{r}
LM_Total_eaten_OP <- lmer(Total_eaten ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_Total_eaten_OP)
```

Vérification des hypothèses
```{r}
qqnorm(resid(LM_Total_eaten_OP))
qqline(resid(LM_Total_eaten_OP))
```
C'est très bon.

Homogénité des variances
```{r}
plot(LM_Total_eaten_OP)
```
c'est correcte

Distance de cook
```{r}
plot(cooks.distance(LM_Total_eaten_OP))
```


### Simplification du modèle


```{r}
drop1(LM_Total_eaten_OP)
```
```{r}
LM_Total_eaten1_OP <- update(LM_Total_eaten_OP, . ~ . - Treatment_bis:Sex)
drop1(LM_Total_eaten1_OP)
```
```{r}
LM_Total_eaten2_OP <- update(LM_Total_eaten1_OP, . ~ . - Treatment_bis:Region)
drop1(LM_Total_eaten2_OP)
```

On arrête la simplification de modèle.

Est-ce qu'elle a été utile ?
```{r}
AIC(LM_Total_eaten2_OP)
AIC(LM_Total_eaten_OP)
```
Non


### Treatment VS Treatment_bis

```{r}
AIC(lmer(Total_eaten ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_Total_eaten_OP))
AIC(lmer(Total_eaten ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment:Sex + Treatment:Region, data = Data_Total_eaten_OP))
```
```{r}
summary(lmer(Total_eaten ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_Total_eaten_OP))$coeff
summary(lmer(Total_eaten ~ Treatment + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment:Sex + Treatment:Region, data = Data_Total_eaten_OP))$coeff
```

Aucun des 2 ne se justifie tellement par rapport à l'autre donc je vais continuer avec treatment_bis qui est plus fréquemment repris dans les autres modèles et plus rigoureux scientifiquement parlant.

### Résultats

Analyse des P-val
```{r}
summary(lmer(Total_eaten ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_Total_eaten_OP))
summary(lmer(Total_eaten ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_Total_eaten_OP_ref_smart))
summary(lmer(Total_eaten ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_Total_eaten_OP_ref_full ))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(lmer(Total_eaten ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_Total_eaten_OP))$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(lmer(Total_eaten ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_Total_eaten_OP_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(lmer(Total_eaten ~ Treatment_bis + Sex + Region + (1 | Mother) + (1 |  Site) + (1 | Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_Total_eaten_OP_ref_full ))$coeff)[,]))

Results_Total_eaten_OP <- rbind(test1,test2,test3)

colnames(Results_Total_eaten_OP ) <- c("Reference","Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_Total_eaten_OP , file = "output/Results_Total_eaten_OP.csv" )

kable(Results_Total_eaten_OP , caption = " Quantité mangée OP ", padding = 1 )
```

```{r}
Info_Total_eaten_OP <- "
Mâles Dark mangent presque significativement moins
"
Info_Total_eaten_OP
```




## Total_eaten AE

### Test

```{r}
LM_Total_eaten_AE <- lmer(Total_eaten ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_AE)
summary(LM_Total_eaten_AE)
```

Vérification des hypothèses
```{r}
qqnorm(resid(LM_Total_eaten_AE))
qqline(resid(LM_Total_eaten_AE))
```
il faudrait faire quelque chose


```{r}
LM_Total_eaten_AE <- lmer(log(Total_eaten) ~ Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region, data = Data_AE)
```

Vérification des hypothèses
```{r}
qqnorm(resid(LM_Total_eaten_AE))
qqline(resid(LM_Total_eaten_AE))
```
C'est très bon.

Homogénité des variances
```{r}
plot(LM_Total_eaten_AE)
```
c'est correcte

Distance de cook
```{r}
plot(cooks.distance(LM_Total_eaten_AE))
```


### Simplification du modèle


```{r}
drop1(LM_Total_eaten_AE)
```
```{r}
LM_Total_eaten1_AE <- update(LM_Total_eaten_AE, . ~ . - Treatment_bis:Sex)
drop1(LM_Total_eaten1_AE)
```

On arrête la simplification de modèle.

Est-ce qu'elle a été utile ?
```{r}
AIC(LM_Total_eaten1_AE)
AIC(LM_Total_eaten_AE)
```
Oui

```{r}
LM_Total_eaten_AE <- LM_Total_eaten1_AE
```



### Treatment VS Treatment_bis

```{r}
AIC(lmer(log(Total_eaten) ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE))
AIC(lmer(log(Total_eaten) ~ Treatment + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment:Region, data = Data_AE))
```
```{r}
summary(lmer(log(Total_eaten) ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE))$coeff
summary(lmer(log(Total_eaten) ~ Treatment + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment:Region, data = Data_AE))$coeff
```

Aucun des 2 ne se justifie tellement par rapport à l'autre donc je vais continuer avec treatment_bis qui est plus fréquemment repris dans les autres modèles et plus rigoureux scientifiquement parlant.


### Résultats

Analyse des P-val
```{r}
summary(lmer(log(Total_eaten) ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE))
summary(lmer(log(Total_eaten) ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE_ref_smart))
summary(lmer(log(Total_eaten) ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE_ref_full))
summary(lmer(log(Total_eaten) ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE_ref_mixt_smart))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(lmer(log(Total_eaten) ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE))$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(lmer(log(Total_eaten) ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(lmer(log(Total_eaten) ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE_ref_full))$coeff)[,]))

test4 <- data.frame(Reference = "Mixt_smart", as.data.frame(( summary(lmer(log(Total_eaten) ~ Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Region, data = Data_AE_ref_mixt_smart))$coeff)[,]))

Results_Total_eaten_AE <- rbind(test1,test2,test3,test4)

colnames(Results_Total_eaten_AE ) <- c("Reference","Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_Total_eaten_AE , file = "output/Results_Total_eaten_AE.csv" )

kable(Results_Total_eaten_AE , caption = " Quantité mangée AE ", padding = 1 )
```

```{r}
Info_Total_eaten_AE <- "
Les individus dans le traitement mixt smart et qui étaient dans la région luxembourgeoise ont mangé significativement plus que l'effet de la région luxembourgeoise et de mixt smart additioné.
"
Info_Total_eaten_AE
```




## Pupa_mass * TT_pupation OP


### Test
```{r}
LM_Pupa_mass_by_TT_pupation_OP <- lmer(Pupa_mass ~ TT_pupation + Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region + TT_pupation:Sex + TT_pupation:Region + TT_pupation:Treatment_bis, data = Data_OP)
summary(LM_Pupa_mass_by_TT_pupation_OP)
```

Test des hypothèses

Normalité des réisuds
```{r}
qqnorm(resid(LM_Pupa_mass_by_TT_pupation_OP))
qqline(resid(LM_Pupa_mass_by_TT_pupation_OP))
```


Homogénité des variances
```{r}
plot(LM_Pupa_mass_by_TT_pupation_OP)
```

Distance de cook
```{r}
plot(cooks.distance(LM_Pupa_mass_by_TT_pupation_OP))
```

### Simplification du modèle

```{r}
drop1(LM_Pupa_mass_by_TT_pupation_OP)
```
```{r}
LM_Pupa_mass_by_TT_pupation1_OP <- update(LM_Pupa_mass_by_TT_pupation_OP, . ~ . - Treatment_bis:Region)
drop1(LM_Pupa_mass_by_TT_pupation1_OP)
```
```{r}
LM_Pupa_mass_by_TT_pupation2_OP <- update(LM_Pupa_mass_by_TT_pupation1_OP, . ~ . - TT_pupation:Treatment_bis)
drop1(LM_Pupa_mass_by_TT_pupation2_OP)
```
```{r}
LM_Pupa_mass_by_TT_pupation3_OP <- update(LM_Pupa_mass_by_TT_pupation2_OP, . ~ . - TT_pupation:Sex)
drop1(LM_Pupa_mass_by_TT_pupation3_OP)
```
```{r}
LM_Pupa_mass_by_TT_pupation4_OP <- update(LM_Pupa_mass_by_TT_pupation3_OP, . ~ . - TT_pupation:Region)
drop1(LM_Pupa_mass_by_TT_pupation4_OP)
```

Il n'y a plus d'interactions qui n'ont pas de P-val significative, nous décidons d'arrêter la simplification du modèle.

Est ce que cela a été utile
```{r}
AIC(LM_Pupa_mass_by_TT_pupation4_OP)
AIC(LM_Pupa_mass_by_TT_pupation_OP)
```
Oui ca a été utile
```{r}
LM_Pupa_mass_by_TT_pupation_OP <- LM_Pupa_mass_by_TT_pupation4_OP
```


### Treatment VS Teatment_bis

```{r}
AIC(lmer(Pupa_mass ~ TT_pupation + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Sex, data = Data_OP))
AIC(lmer(Pupa_mass ~ TT_pupation + Treatment + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment:Sex, data = Data_OP))
```
```{r}
summary(lmer(Pupa_mass ~ TT_pupation + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Sex, data = Data_OP))$coeff
summary(lmer(Pupa_mass ~ TT_pupation + Treatment + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment:Sex, data = Data_OP))$coeff
```

Rien de très différent, on peut garder Treatment qui serait mieux selon l'AIC

### Résultats

Analyse des pval
```{r}
summary(lmer(Pupa_mass ~ TT_pupation + Treatment + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment:Sex, data = Data_OP))
summary(lmer(Pupa_mass ~ TT_pupation + Treatment + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment:Sex, data = Data_OP_ref_full))
summary(lmer(Pupa_mass ~ TT_pupation + Treatment + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment:Sex, data = Data_OP_ref_smart))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(lmer(Pupa_mass ~ TT_pupation + Treatment + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment:Sex, data = Data_OP))$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(lmer(Pupa_mass ~ TT_pupation + Treatment + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment:Sex, data = Data_OP_ref_full))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(lmer(Pupa_mass ~ TT_pupation + Treatment + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment:Sex, data = Data_OP_ref_smart))$coeff)[,]))

Results_Pupa_mass_by_TT_pupation_OP <- rbind(test1,test2,test3)

colnames(Results_Pupa_mass_by_TT_pupation_OP ) <- c("Reference","Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_Pupa_mass_by_TT_pupation_OP , file = "output/Results_Pupa_mass_by_TT_pupation_OP.csv" )

kable(Results_Pupa_mass_by_TT_pupation_OP , caption = " Poids des chrysalides en fonction du temps pour rentrer en chrysalide OP", padding = 1 )
```

```{r} 
Info_Pupa_mass_by_TT_pupation_OP <- "
Au plus un individu met du temps pour entrer en chrysaldie, au plus sa chrysaldie est lourde (ce n'est pas parfaitement significatif p-val = 0.0506)
Les individus en traitement Full ont quasiment significativement (p-val = 0.0684) des chrysalides plus lourdes (0.0083 g) que les inbdividus en traitement dark. 
Les chrysalides des individus mâles du traitement Smart sont significativements (p-val = 0.0235) plus lourdes (0.011 g) que celles des femelles.
"
Info_Pupa_mass_by_TT_pupation_OP
```



## Pupa_mass * TT_pupation AE


### Test
```{r}
LM_Pupa_mass_by_TT_pupation_AE <- lmer(Pupa_mass ~ TT_pupation + Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region + TT_pupation:Sex + TT_pupation:Region + TT_pupation:Treatment_bis, data = Data_AE)
summary(LM_Pupa_mass_by_TT_pupation_AE)
```

Test des hypothèses

Normalité des réisuds
```{r}
qqnorm(resid(LM_Pupa_mass_by_TT_pupation_AE))
qqline(resid(LM_Pupa_mass_by_TT_pupation_AE))
```


Homogénité des variances
```{r}
plot(LM_Pupa_mass_by_TT_pupation_AE)
```

Distance de cook
```{r}
plot(cooks.distance(LM_Pupa_mass_by_TT_pupation_AE))
```

1 gros outlier à retirer si possible
```{r}
Data_Pupa_mass_AE <- (Data_AE %>% filter(Sex != "NA", Pupa_mass != "NA" ))[unname(cooks.distance(LM_Pupa_mass_by_TT_pupation_AE) != max(cooks.distance(LM_Pupa_mass_by_TT_pupation_AE))),]

Data_Pupa_mass_AE_ref_smart <- Data_Pupa_mass_AE
Data_Pupa_mass_AE_ref_smart$Treatment_bis <- relevel(Data_Pupa_mass_AE$Treatment_bis, ref = "Smart")

Data_Pupa_mass_AE_ref_full <- Data_Pupa_mass_AE
Data_Pupa_mass_AE_ref_full$Treatment_bis <- relevel(Data_Pupa_mass_AE$Treatment_bis, ref = "Full")

Data_Pupa_mass_AE_ref_mixt_smart <- Data_Pupa_mass_AE
Data_Pupa_mass_AE_ref_mixt_smart$Treatment_bis <- relevel(Data_Pupa_mass_AE$Treatment_bis, ref = "Mixt_smart")
```


Et on refait les tests

```{r}
LM_Pupa_mass_by_TT_pupation_AE <- lmer(Pupa_mass ~ TT_pupation + Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region + TT_pupation:Sex + TT_pupation:Region + TT_pupation:Treatment_bis, data = Data_Pupa_mass_AE)
```

Test des hypothèses

Normalité des réisuds
```{r}
qqnorm(resid(LM_Pupa_mass_by_TT_pupation_AE))
qqline(resid(LM_Pupa_mass_by_TT_pupation_AE))
```


Homogénité des variances
```{r}
plot(LM_Pupa_mass_by_TT_pupation_AE)
```

Distance de cook
```{r}
plot(cooks.distance(LM_Pupa_mass_by_TT_pupation_AE))
```

### Simplification du modèle

```{r}
drop1(LM_Pupa_mass_by_TT_pupation_AE)
```
```{r}
LM_Pupa_mass_by_TT_pupation1_AE <- update(LM_Pupa_mass_by_TT_pupation_AE, . ~ . - TT_pupation:Treatment_bis)
drop1(LM_Pupa_mass_by_TT_pupation1_AE)
```
```{r}
LM_Pupa_mass_by_TT_pupation2_AE <- update(LM_Pupa_mass_by_TT_pupation1_AE, . ~ . - Treatment_bis:Region)
drop1(LM_Pupa_mass_by_TT_pupation2_AE)
```
```{r}
LM_Pupa_mass_by_TT_pupation3_AE <- update(LM_Pupa_mass_by_TT_pupation2_AE, . ~ . - TT_pupation:Sex)
drop1(LM_Pupa_mass_by_TT_pupation3_AE)
```


Il n'y a plus d'interactions qui n'ont pas de P-val significative, nous décidons d'arrêter la simplification du modèle.

Est ce que cela a été utile
```{r}
AIC(LM_Pupa_mass_by_TT_pupation3_AE)
AIC(LM_Pupa_mass_by_TT_pupation_AE)
```
Oui ca a été utile
```{r}
LM_Pupa_mass_by_TT_pupation_AE <- LM_Pupa_mass_by_TT_pupation3_AE
```


### Treatment VS Teatment_bis

```{r}
AIC(lmer(Pupa_mass ~ TT_pupation + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Sex + TT_pupation:Region, data = Data_Pupa_mass_AE))
AIC(lmer(Pupa_mass ~ TT_pupation + Treatment + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment:Sex + TT_pupation:Region, data = Data_Pupa_mass_AE))
```
```{r}
summary(lmer(Pupa_mass ~ TT_pupation + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Sex + TT_pupation:Region, data = Data_Pupa_mass_AE))$coeff
summary(lmer(Pupa_mass ~ TT_pupation + Treatment + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment:Sex + TT_pupation:Region, data = Data_Pupa_mass_AE))$coeff
```
Même si l'AIC trouve que c'est mieux de prendre treatment, comme Mixt smart a un effet significatif, je compte prendre Treatment_bis

### Résultats

Analyse des pval
```{r}
summary(lmer(Pupa_mass ~ TT_pupation + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Sex + TT_pupation:Region, data = Data_Pupa_mass_AE))
summary(lmer(Pupa_mass ~ TT_pupation + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Sex + TT_pupation:Region, data = Data_Pupa_mass_AE_ref_smart))
summary(lmer(Pupa_mass ~ TT_pupation + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Sex + TT_pupation:Region, data = Data_Pupa_mass_AE_ref_full))
summary(lmer(Pupa_mass ~ TT_pupation + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Sex + TT_pupation:Region, data = Data_Pupa_mass_AE_ref_mixt_smart))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(lmer(Pupa_mass ~ TT_pupation + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Sex + TT_pupation:Region, data = Data_Pupa_mass_AE))$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(lmer(Pupa_mass ~ TT_pupation + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Sex + TT_pupation:Region, data = Data_Pupa_mass_AE_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(lmer(Pupa_mass ~ TT_pupation + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Sex + TT_pupation:Region, data = Data_Pupa_mass_AE_ref_full))$coeff)[,]))

test4 <- data.frame(Reference = "Mixt_smart", as.data.frame(( summary(lmer(Pupa_mass ~ TT_pupation + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Sex + TT_pupation:Region, data = Data_Pupa_mass_AE_ref_mixt_smart))$coeff)[,]))

Results_Pupa_mass_by_TT_pupation_AE <- rbind(test1,test2,test3,test4)

colnames(Results_Pupa_mass_by_TT_pupation_AE ) <- c("Reference","Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_Pupa_mass_by_TT_pupation_AE , file = "output/Results_Pupa_mass_by_TT_pupation_AE.csv" )

kable(Results_Pupa_mass_by_TT_pupation_AE , caption = " Poids des chrysalides en fonction du temps pour rentrer en chrysalide AE", padding = 1 )
```


```{r} 
Info_Pupa_mass_by_TT_pupation_AE <- "
Au plus un individu met du temps pour entrer en chrysalide, au plus sa chrysalide sera lourde (p-val = 6.90e-12).
Les chrysalides des individus des traitements Full et Smart sont significatiment (p-val = 0.0409 et 0.0159 respectivement) plus lourdes (0.027 g et 0.036 g respectivement) que celles du traitement Dark.
Celles de Mixt-smart sont significativement (p-val = 1.14e-05 et 0.00447 et 0.0328 respectivement) plus lourdes (0.073 g, 0.046 g et 0.038 g respectivement) que celles des traitements Dark, Full et Smart.
Les chrysalides venant d'individus mâles dans le traitement Dark sont significativement (p-val = 0.0198) plus lourdes (0.034 g) que celles des femelles.
Les chrysalides des individus déscendant d'une mère provenant de la région luxembourgeoise sont quasiement significativement (p-val = 0.0539) plus lourdes (0.069 g) que celles du brabant mais les individus ont significativement (p-val = 0.0481) moins bien optimisé le temps avant l'entrée en chrysaldie pour augmenter la masse finale de la chrysalide (0.0014 g en moins par semaine passée).
"
Info_Pupa_mass_by_TT_pupation_AE
```





## Mass_by_Time OP

Dévelloppement du jeux de données
```{r}
Data_mass <- Data[,1:31] %>% pivot_longer(cols = M0:M20, names_to = "Numéro_de_masse", values_to = "Mass")
Data_mass$Time <- rep(0:20, length.out = nrow(Data_mass))
```

```{r}
Data_mass_OP <- Data_mass %>% filter(Species == "OP")
Data_mass_AE <- Data_mass %>% filter(Species == "AE")
```


```{r}
Data_mass_OP_ref_smart <- Data_mass_OP
Data_mass_OP_ref_smart$Treatment <- relevel(Data_mass_OP$Treatment, ref = "Smart")
Data_mass_OP_ref_smart$Treatment_bis <- relevel(Data_mass_OP$Treatment_bis, ref = "Smart")

Data_mass_OP_ref_full <- Data_mass_OP
Data_mass_OP_ref_full$Treatment <- relevel(Data_mass_OP$Treatment, ref = "Full")
Data_mass_OP_ref_full$Treatment_bis <- relevel(Data_mass_OP$Treatment_bis, ref = "Full")

Data_mass_AE_ref_smart <- Data_mass_AE
Data_mass_AE_ref_smart$Treatment <- relevel(Data_mass_AE$Treatment, ref = "Smart")
Data_mass_AE_ref_smart$Treatment_bis <- relevel(Data_mass_AE$Treatment_bis, ref = "Smart")

Data_mass_AE_ref_full <- Data_mass_AE
Data_mass_AE_ref_full$Treatment <- relevel(Data_mass_AE$Treatment, ref = "Full")
Data_mass_AE_ref_full$Treatment_bis <- relevel(Data_mass_AE$Treatment_bis, ref = "Full")

Data_mass_AE_ref_mixt_smart <- Data_mass_AE
Data_mass_AE_ref_mixt_smart$Treatment_bis <- relevel(Data_mass_AE$Treatment_bis, ref = "Mixt_smart")
```


### Test

```{r}
LM_mass_by_time_OP <- lmer(Mass ~ Time + Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region + Time:Sex + Time:Region + Time:Treatment_bis, data = Data_mass_OP)
summary(LM_mass_by_time_OP)
```

Vérification des hypothèses

```{r}
qqnorm(resid(LM_mass_by_time_OP))
qqline(resid(LM_mass_by_time_OP))
```
Il faudrait modifier les données

```{r}
LM_mass_by_time_OP <- lmer(sqrt(Mass) ~ Time + Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region + Time:Sex + Time:Region + Time:Treatment_bis, data = Data_mass_OP)
```

```{r}
qqnorm(resid(LM_mass_by_time_OP))
qqline(resid(LM_mass_by_time_OP))
```

Homogénité des variances
```{r}
plot(LM_mass_by_time_OP)
```

Distance de cook
```{r}
plot(cooks.distance(LM_mass_by_time_OP))
```

### Simplification du modèle

```{r}
drop1(LM_mass_by_time_OP)
```
```{r}
LM_mass_by_time1_OP <- update(LM_mass_by_time_OP, . ~ . - Time:Sex)
drop1(LM_mass_by_time1_OP)
```
```{r}
LM_mass_by_time2_OP <- update(LM_mass_by_time1_OP, . ~ . - Treatment_bis:Region)
drop1(LM_mass_by_time2_OP)
```
```{r}
LM_mass_by_time3_OP <- update(LM_mass_by_time2_OP, . ~ . - Time:Treatment_bis)
drop1(LM_mass_by_time3_OP)
```
```{r}
LM_mass_by_time4_OP <- update(LM_mass_by_time3_OP, . ~ . - Time:Region)
drop1(LM_mass_by_time4_OP)
```
Plus aucune interaction ne peut être retirée, on arrête la simplification ici.

A-t-elle été utile ?
```{r}
AIC(LM_mass_by_time4_OP)
AIC(LM_mass_by_time_OP)
```
Oui

```{r}
LM_mass_by_time_OP <- LM_mass_by_time4_OP
```


### Treatment VS Teatment_bis

```{r}
AIC(lmer(sqrt(Mass) ~ Time + Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Sex, data = Data_mass_OP))
AIC(lmer(sqrt(Mass) ~ Time + Treatment + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment:Sex, data = Data_mass_OP))
```
```{r}
summary(lmer(sqrt(Mass) ~ Time + Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Sex, data = Data_mass_OP))$coeff
summary(lmer(sqrt(Mass) ~ Time + Treatment + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment:Sex, data = Data_mass_OP))$coeff
```
Si je me base sur les AIC je continue avec treatment_bis

### Résultats

Analyse des P-val
```{r}
summary(lmer(sqrt(Mass) ~ Time + Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Sex, data = Data_mass_OP))
summary(lmer(sqrt(Mass) ~ Time + Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Sex, data = Data_mass_OP_ref_full))
summary(lmer(sqrt(Mass) ~ Time + Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Sex, data = Data_mass_OP_ref_smart))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(lmer(sqrt(Mass) ~ Time + Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Sex, data = Data_mass_OP))$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(lmer(sqrt(Mass) ~ Time + Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Sex, data = Data_mass_OP_ref_full))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(lmer(sqrt(Mass) ~ Time + Treatment_bis + Sex + Region + (1 | Mother) +  (1 | Site) + (1 | Room) + Treatment_bis:Sex, data = Data_mass_OP_ref_smart))$coeff)[,]))

Results_Mass_by_time_OP <- rbind(test1,test2,test3)

colnames(Results_Mass_by_time_OP ) <- c("Reference","Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_Mass_by_time_OP , file = "output/Results_Mass_by_time_OP.csv" )

kable(Results_Mass_by_time_OP , caption = " Poids en fonction du temps OP ", padding = 1 )
```

```{r}
Info_Mass_by_Time_OP <- "
Augmentation de la masse avec le temps.
Smart est plus léger que Dark.
Full est presque plus lourd que smart (p val 0.0849)
"
Info_Mass_by_Time_OP
```




## Mass_by_Time AE

### Test

```{r}
LM_mass_by_time_AE <- lmer(Mass ~ Time + Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region + Time:Sex + Time:Region + Time:Treatment_bis, data = Data_mass_AE)
summary(LM_mass_by_time_AE)
```

Vérification des hypothèses

```{r}
qqnorm(resid(LM_mass_by_time_AE))
qqline(resid(LM_mass_by_time_AE))
```
Il faudrait modifier les données

```{r}
LM_mass_by_time_AE <- lmer(log(Mass) ~ log(Time + 1) + Treatment_bis + Sex + Region + (1|Mother) + (1|Site) + (1|Room) + Treatment_bis:Sex + Treatment_bis:Region + Time:Sex + Time:Region + Time:Treatment_bis, data = Data_mass_AE)
```

```{r}
qqnorm(resid(LM_mass_by_time_AE))
qqline(resid(LM_mass_by_time_AE))
```

Homogénité des variances
```{r}
plot(LM_mass_by_time_AE)
```

Distance de cook
```{r}
plot(cooks.distance(LM_mass_by_time_AE))
```

### Simplification du modèle

```{r}
drop1(LM_mass_by_time_AE)
```
```{r}
LM_mass_by_time1_AE <- update(LM_mass_by_time_AE, . ~ . - Region:Time)
drop1(LM_mass_by_time1_AE)
```
```{r}
LM_mass_by_time2_AE <- update(LM_mass_by_time1_AE, . ~ . - Treatment_bis:Sex)
drop1(LM_mass_by_time2_AE)
```

Plus aucune interaction ne peut être retirée, on arrête la simplification ici.

A-t-elle été utile ?
```{r}
AIC(LM_mass_by_time2_AE)
AIC(LM_mass_by_time_AE)
```
Oui

```{r}
LM_mass_by_time_AE <- LM_mass_by_time2_AE
```


### Treatment VS Teatment_bis

```{r}
AIC(lmer(log(Mass) ~ log(Time + 1) + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Sex:Time + Treatment_bis:Time, data = Data_mass_AE))
AIC(lmer(log(Mass) ~ log(Time + 1) + Treatment + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment:Region +  Sex:Time + Treatment:Time, data = Data_mass_AE))
```
```{r}
summary(lmer(log(Mass) ~ log(Time + 1) + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Sex:Time + Treatment_bis:Time, data = Data_mass_AE))$coeff
summary(lmer(log(Mass) ~ log(Time + 1) + Treatment + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment:Region +  Sex:Time + Treatment:Time, data = Data_mass_AE))$coeff
```
Je continue avec treatment_bis

### Résultats

Analyse des P-val
```{r}
summary(lmer(log(Mass) ~ log(Time + 1) + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Sex:Time + Treatment_bis:Time, data = Data_mass_AE))
summary(lmer(log(Mass) ~ log(Time + 1) + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Sex:Time + Treatment_bis:Time, data = Data_mass_AE_ref_smart))
summary(lmer(log(Mass) ~ log(Time + 1) + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Sex:Time + Treatment_bis:Time, data = Data_mass_AE_ref_full))
summary(lmer(log(Mass) ~ log(Time + 1) + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Sex:Time + Treatment_bis:Time, data = Data_mass_AE_ref_mixt_smart))
```

```{r}
test1 <- data.frame(Reference = "Dark", as.data.frame(( summary(lmer(log(Mass) ~ log(Time + 1) + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Sex:Time + Treatment_bis:Time, data = Data_mass_AE))$coeff)[,]))

test2 <- data.frame(Reference = "Smart", as.data.frame(( summary(lmer(log(Mass) ~ log(Time + 1) + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Sex:Time + Treatment_bis:Time, data = Data_mass_AE_ref_smart))$coeff)[,]))

test3 <- data.frame(Reference = "Full", as.data.frame(( summary(lmer(log(Mass) ~ log(Time + 1) + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Sex:Time + Treatment_bis:Time, data = Data_mass_AE_ref_full))$coeff)[,]))

test4 <- data.frame(Reference = "Mixt_smart", as.data.frame(( summary(lmer(log(Mass) ~ log(Time + 1) + Treatment_bis + Sex + Region + (1 |  Mother) + (1 | Site) + (1 | Room) + Treatment_bis:Region +  Sex:Time + Treatment_bis:Time, data = Data_mass_AE_ref_mixt_smart))$coeff)[,]))

Results_Mass_by_time_AE <- rbind(test1,test2,test3,test4)

colnames(Results_Mass_by_time_AE ) <- c("Reference","Estimate", "Std. Error", "df", "t.value", "Pr(>|z|)")

write.csv2( Results_Mass_by_time_AE , file = "output/Results_Mass_by_time_AE.csv" )

kable(Results_Mass_by_time_AE , caption = " Poids en fonction du temps OP ", padding = 1 )
```

```{r}
Info_Mass_by_Time_AE <- "
Augmentation de la masse en fonction du temps
Smart est plus légere que n'importe quel autre traitement
Luxembourg est plus léger que Brabant pour les individus dans Dark
Luxembourg est plus léger que Brabant pour les individus dans Full
Par ordre d'optimisation du temps pour prendre de la masse croissant : Dark < Full & Mixt Smart < Smart
"
Info_Mass_by_Time_AE
```