### Objective ###

" The objective of this script is to analyse together american and european
cline data using a binomial law. We also represent the sampling sites on a map
colored by climatic variables. We also test for the HW equilibrium in every station"


### Methods ###




### Packages ###
library(raster)         # Manipulation des ratsers
library(geodata)
library(sf)             # Données géospatiales
library(ggplot2)        # Représentation
library(rnaturalearth)
library(rnaturalearthdata)
library(terra)
library(patchwork)
library(betareg)
library(RVAideMemoire)
library(car)
library(dplyr)
library(xml2)
library(MuMIn)
library(HardyWeinberg)



### Working directory ###
setwd("S:/Doctorat/Communication/Articles/Premier article/Article/VF reviewed 4/Supplementary")


################################################################################

### Import data ###

# GPS points
location_GPS<- read.delim("Data_clines.txt")
location_GPS = location_GPS[1:26,]
location_GPS$Continent = as.factor(location_GPS$Continent)
location_GPS$Population = as.factor(location_GPS$Population)

# Temperature raster
# temp = raster("H:/Doctorat/1 Distribution et Phénotypes/1B Clines/Analyses/wc2.1_10m_bio/wc2.1_10m_bio_1.tif")
temp = raster("S:/Doctorat/1 Distribution et Phénotypes/1B Clines/Analyses/wc2.1_2.5m_bio/wc2.1_2.5m_bio_1.tif")

# BIO1 = Annual Mean Temperature
# BIO2 = Mean Diurnal Range (Mean of monthly (max temp - min temp))
# BIO3 = Isothermality (BIO2/BIO7) (×100)
# BIO4 = Temperature Seasonality (standard deviation ×100)
# BIO5 = Max Temperature of Warmest Month
# BIO6 = Min Temperature of Coldest Month
# BIO7 = Temperature Annual Range (BIO5-BIO6)
# BIO8 = Mean Temperature of Wettest Quarter
# BIO9 = Mean Temperature of Driest Quarter
# BIO10 = Mean Temperature of Warmest Quarter
# BIO11 = Mean Temperature of Coldest Quarter
# BIO12 = Annual PrecipitationTemperature monitoring device
# BIO13 = Precipitation of Wettest Month
# BIO14 = Precipitation of Driest Month
# BIO15 = Precipitation Seasonality (Coefficient of Variation)
# BIO16 = Precipitation of Wettest Quarter
# BIO17 = Precipitation of Driest Quarter
# BIO18 = Precipitation of Warmest Quarter
# BIO19 = Precipitation of Coldest Quarter

"We further derived 10 bioclimatic variables describing the annual mean, range,
variance, and extreme values for temperature and salinity from long-term monthly
climatological means obtained from remotely sensed and in situ oceanographic
observations."

# temp = raster("S:/Doctorat/1 Distribution et Phénotypes/1B Clines/Analyses/Sea temperature/biogeo08_17_30s/biogeo15_30s/w001000.adf")
# aoi = extent(-75,20,40,70)
# options(scipen = 0)
# temp = crop(temp, aoi)
# temp = aggregate(temp, fact = 5) # Divise la résolution par 5


# Transformation en un objet maniupulable dans ggplot2
#temp <- crop(temp, extent(-7, 20, 40, 70))
temp_df <- as.data.frame(temp, xy = TRUE, na.rm = TRUE)
colnames(temp_df) <- c("longitude", "latitude", "temperature")

# Sélection de la zone d'intérêt
#temp_df = subset(temp_df, temp_df$longitude > -10 & temp_df$longitude < 20 & temp_df$latitude > 45 & temp_df$latitude < 70)
temp_df = subset(temp_df, temp_df$longitude > -7 & temp_df$longitude < 20 & temp_df$latitude > 40 & temp_df$latitude < 70)


### Traçage carte de températures ###

# ggplot() +
# geom_raster(data = temp_df, aes(x = longitude, y = latitude, fill = temperature)) + # Tracer les températures (fond coloré)
# scale_fill_gradientn(colors = c("blue", "green", "yellow", "red"), name = "Temperature (°C)") + # Personnaliser la palette de couleurs
# labs(title = "Annual mean temperatures", x = "Longitude", y = "Latitude") +
# theme_minimal()

ggplot() +
  geom_tile(data = temp_df, aes(x = longitude, y = latitude, fill = temperature)) + # Tracer les températures (fond coloré)
  scale_fill_gradientn(colors = c("dodgerblue4", "chartreuse3", "gold", "orangered3"), name = "Temperature (°C)") + # Personnaliser la palette de couleurs
  labs(title = "Mean temperature of the coldest month", x = "Longitude", y = "Latitude") +
  theme_minimal()






### Ajout des sites d'échantillonnage ###

# Conversion des sites d'échantillonnage en un objet spatial
points_sf <- st_as_sf(location_GPS, coords = c("Longitude", "Latitude"), crs = 4326)
points_sf$label <- paste(points_sf$Population, points_sf$Date, sep = "  ")  # Saut de ligne entre nom et date
st_crs(points_sf) # Vérification que le spoints sont dans le même système de projection que le raster

# Traçage du plot
plot_map = ggplot() +
  geom_raster(data = temp_df, aes(x = longitude, y = latitude, fill = temperature)) +
  scale_fill_gradientn(colors = c("dodgerblue4", "chartreuse3", "gold", "orangered3"), name = "Temperature (°C)") +
  geom_sf(data = subset(points_sf, points_sf$Continent=="E"), size = 3, shape = 1, stroke = 1.3) +
  #geom_text(data = subset(points_sf, points_sf$Continent=="E"), aes(x = st_coordinates(geometry)[, 1], y = st_coordinates(geometry)[, 2], label = label), hjust = -0.2, vjust = -0.2, size = 3) +
  labs(title = "Annual mean temperatures", x = "Longitude", y = "Latitude") +
  theme_minimal() +
  theme(
    axis.text.x = element_text(size = 14),
    axis.text.y = element_text(size = 14), 
    legend.position = "none",  # Position en haut à gauche (x = gauche/droite, y = haut/bas)
  )

plot_map

# Extraire les valeurs du raster aux positions des points
values <- extract(temp, st_coordinates(points_sf))

# Ajouter les valeurs extraites au data frame des points
points_sf$valeur_raster <- values
location_GPS$Temperature <- values
print(points_sf)


################################################################################

# Plot latitude on A_freq :

plot_lat = ggplot(location_GPS) +
  aes(x=Freq_A, y=Latitude) +
  geom_point(size = 4) +
  xlim(0,1) +
  theme_classic() +
  geom_text(label = location_GPS$Population, vjust = -1, hjust = 1)


# Plot temperature on A_freq

plot_temp = ggplot(location_GPS) +
  aes(x=Freq_A, y=Temperature) +
  geom_point(size = 4) +
  xlim(0,1) +
  theme_classic() +
  geom_text(label = location_GPS$Population, vjust = -1, hjust = 1)

plot_lat + plot_temp


################################################################################

# Deux continents combinés
modT1 = glm(cbind(N_A,N_B) ~ Latitude, data=location_GPS, family=binomial(link="logit"), na.action = "na.pass")
plotresid(modT1)
summary(modT1)
Anova(modT1)
r.squaredGLMM(modT1) # 0.74

modT2 = glm(cbind(N_A,N_B) ~ Temperature, data=location_GPS, family=binomial(link="logit"), na.action = "na.pass")
plotresid(modT2)
summary(modT2)
Anova(modT2)
r.squaredGLMM(modT2)


# Continent européen
modE1 = glm(cbind(N_A,N_B) ~ Latitude, data = subset(location_GPS, location_GPS$Continent == "E"), family=binomial(link="logit"), na.action = "na.pass")
plotresid(modE1)
Anova(modE1)
summary(modE1)
r.squaredGLMM(modE1)

modE2 = glm(cbind(N_A,N_B) ~ Temperature, data = subset(location_GPS, location_GPS$Continent == "E"), family=binomial(link="logit"), na.action = "na.pass")
plotresid(modE2)
Anova(modE2)
summary(modE2)
r.squaredGLMM(modE2)

# Continent américain
modA1 = glm(cbind(N_A,N_B) ~ Latitude, data = subset(location_GPS, location_GPS$Continent == "A"), family=binomial(link="logit"), na.action = "na.pass")
plotresid(modA1)
Anova(modA1)
summary(modA1)
r.squaredGLMM(modA1)


modA2 = glm(cbind(N_A,N_B) ~ Temperature, data = subset(location_GPS, location_GPS$Continent == "A"), family=binomial(link="logit"), na.action = "na.pass")
plotresid(modA2)
Anova(modA2)
summary(modA2)
r.squaredGLMM(modA2)


################################################################################

# Réprésentation des regressions en fonction de la latitude


# Tableau des valeurs à prédire
to_predict = seq(40, 70, 0.01)
len = length(to_predict)
predict_data = as.data.frame(matrix(data = 0, nrow = 2*len, ncol = 2))
colnames(predict_data) = c("Continent", "Latitude")
predict_data$Continent[1:len] = "E"
predict_data$Continent[(len+1):(2*len)] = "A"
predict_data$Latitude[1:len] = to_predict
predict_data$Latitude[(len+1):(2*len)] = to_predict

# Calcul des valeurs prédites
predictions_E = predict(modE1, newdata = predict_data[predict_data$Continent == "E", ], type = "response", se.fit = T)
predictions_A = predict(modA1, newdata = predict_data[predict_data$Continent == "A", ], type = "response", se.fit = T)
# predictions_T = predict(modT1, newdata = predict_data, type = "response")

# Remplissage tableau Europe
predict_data$predicted[predict_data$Continent == "E"] <- predictions_E$fit
predict_data$lwr[predict_data$Continent == "E"] <- predictions_E$fit - 1.96 * predictions_E$se.fit
predict_data$upr[predict_data$Continent == "E"] <- predictions_E$fit + 1.96 * predictions_E$se.fit

# Remplissage tableau Amérique
predict_data$predicted[predict_data$Continent == "A"] <- predictions_A$fit
predict_data$lwr[predict_data$Continent == "A"] <- predictions_A$fit - 1.96 * predictions_A$se.fit
predict_data$upr[predict_data$Continent == "A"] <- predictions_A$fit + 1.96 * predictions_A$se.fit

# Remplissage tableaux combinés
# predict_data$predictedT <- predictions_T$fit
# predict_data$T_lwr = predictions_T$fit - 1.96 * predictions_T$se.fit
# predict_data$T_upr = predictions_T$fit + 1.96 * predictions_T$se.fit

# Traçage des courbes
plot_lat2 = ggplot() +
  geom_point(data = subset(location_GPS, is.na(location_GPS$Freq_A) == F), aes(x = Latitude, y = Freq_A, color = Continent), size = 3) +  # Points de données
  geom_point(data = predict_data, aes(y = predicted, x = Latitude, color = Continent), size = 0.5) +  # Courbe de régression pour les deux continents
  # geom_point(data = predict_data, aes(y = predictedT, x = Latitude, color = "black"), size = 0.5) +
  geom_ribbon(data = predict_data[predict_data$Continent == "E", ], aes(x = Latitude, ymin = lwr, ymax = upr), alpha = 0.2) +
  geom_ribbon(data = predict_data[predict_data$Continent == "A", ], aes(x = Latitude, ymin = lwr, ymax = upr), alpha = 0.2) +
  labs(title = "Binomial regression of inversion frequency on latitude",
       x = "Latitude", y = "Frequency of arrangement A") +  # Titres
  scale_color_manual(values = c("E" = "steelblue3", "A" = "indianred3")) +  # Couleurs par continent
  scale_y_continuous(limits = c(0, 1)) +
  scale_x_continuous(limits = c(40, 70)) +
  theme_minimal()+  # Thème minimal
  theme(panel.grid = element_blank(), legend.position = "none") +
  coord_flip()
  # geom_text(data = subset(location_GPS, is.na(location_GPS$Freq_A) == F),
  #           aes(x = Latitude, y = Freq_A, label = Population),
  #           size = 3, hjust = -0.1, vjust = 0)

plot_lat2


################################################################################



# Réprésentation des regressions en fonction de la température


# Tableau des valeurs à prédire
to_predict = seq(0, 12.5, 0.01)
len = length(to_predict)
predict_data = as.data.frame(matrix(data = 0, nrow = 2*len, ncol = 2))
colnames(predict_data) = c("Continent", "Temperature")
predict_data$Continent[1:len] = "E"
predict_data$Continent[(len+1):(2*len)] = "A"
predict_data$Temperature[1:len] = to_predict
predict_data$Temperature[(len+1):(2*len)] = to_predict

# Calcul des valeurs prédites
predictions_E = predict(modE2, newdata = predict_data[predict_data$Continent == "E", ], type = "response", se.fit = T)
predictions_A = predict(modA2, newdata = predict_data[predict_data$Continent == "A", ], type = "response", se.fit = T)
# predictions_T = predict(modT1, newdata = predict_data, type = "response")

# Remplissage tableau Europe
predict_data$predicted[predict_data$Continent == "E"] <- predictions_E$fit
predict_data$lwr[predict_data$Continent == "E"] <- predictions_E$fit - 1.96 * predictions_E$se.fit
predict_data$upr[predict_data$Continent == "E"] <- predictions_E$fit + 1.96 * predictions_E$se.fit

# Remplissage tableau Amérique
predict_data$predicted[predict_data$Continent == "A"] <- predictions_A$fit
predict_data$lwr[predict_data$Continent == "A"] <- predictions_A$fit - 1.96 * predictions_A$se.fit
predict_data$upr[predict_data$Continent == "A"] <- predictions_A$fit + 1.96 * predictions_A$se.fit

# Remplissage tableaux combinés
# predict_data$predictedT <- predictions_T$fit
# predict_data$T_lwr = predictions_T$fit - 1.96 * predictions_T$se.fit
# predict_data$T_upr = predictions_T$fit + 1.96 * predictions_T$se.fit

# Traçage des courbes
plot_temp2 = ggplot() +
  geom_point(data = subset(location_GPS, is.na(location_GPS$Freq_A) == F), aes(x = Temperature, y = Freq_A, color = Continent), size = 3) +  # Points de données
  geom_point(data = predict_data, aes(y = predicted, x = Temperature, color = Continent), size = 0.5) +  # Courbe de régression pour les deux continents
  # geom_point(data = predict_data, aes(y = predictedT, x = Temperature, color = "black"), size = 0.5) +
  geom_ribbon(data = predict_data[predict_data$Continent == "E", ], aes(x = Temperature, ymin = lwr, ymax = upr), alpha = 0.2) +
  geom_ribbon(data = predict_data[predict_data$Continent == "A", ], aes(x = Temperature, ymin = lwr, ymax = upr), alpha = 0.2) +
  labs(title = "Binomial regression of inversion frequency on temperature",
       x = "Annual mean air temperature", y = "Frequency of arrangement A") +  # Titres
  scale_color_manual(values = c("E" = "steelblue3", "A" = "indianred3")) +  # Couleurs par continent
  scale_y_continuous(limits = c(0, 1)) +
  #scale_x_continuous(limits = c(-21, 5)) +
  theme_minimal()+  # Thème minimal
  theme(panel.grid = element_blank(), legend.position = "none") +
  coord_flip() +
  scale_x_reverse()
  # geom_text(data = subset(location_GPS, is.na(location_GPS$Freq_A) == F),
  #           aes(x = Temperature, y = Freq_A, label = Population),
  #           size = 3, hjust = -0.1, vjust = 0)

plot(plot_temp2)


plot_map + plot_lat2 + plot_temp2

ggsave("Clines.tiff", dpi = 600, width = 25, height = 10, units = "cm")



################################################################################

#### Testing Hardy-Weinberg equilibrium in all european stations ####

location_EU = subset(location_GPS, location_GPS$Continent == "E")
location_AM = subset(location_GPS, location_GPS$Continent == "A")

for (i in seq(1,length(location_AM$Population))) {
  x1 = location_AM[i, 10:12]
  x1 = as.numeric(x1)
  print(location_AM[i,2])
  names(x1) = c("AA", "AB", "BB")
  HWAlltests(x1, verbose=T)
}

# FK : OK
# LA : OK
# EG : OK
# OB : bof, à la limite
# SK : OK
# OS : OK
# TJ : OK
# ST : OK
# RO : OK
# VE : pas à l'équilibre





























to_predict = seq(0, 12, 0.01)
len = length(to_predict)

predict_data = as.data.frame(matrix(data = 0, nrow = 2*len, ncol = 2))
colnames(predict_data) = c("Continent", "Temperature")
predict_data$Continent[1:len] = "E"
predict_data$Continent[(len+1):(2*len)] = "A"
predict_data$Temperature[1:len] = to_predict
predict_data$Temperature[(len+1):(2*len)] =to_predict

predict_data$predicted[predict_data$Continent == "E"] <- predict(modE2, newdata = predict_data[predict_data$Continent == "E", ], type = "response")
predict_data$predicted[predict_data$Continent == "A"] = predict(modA2, newdata = predict_data[predict_data$Continent == "A", ], type = "response")
predict_data$predictedT <- predict(modT2, newdata = predict_data, type = "response")



plot_temp2 = ggplot() +
  geom_point(data = subset(location_GPS, is.na(location_GPS$Freq_A) == F), aes(x = Temperature, y = Freq_A, color = Continent), size = 3) +  # Points de données
  geom_point(data = predict_data, aes(y = predicted, x = Temperature, color = Continent), size = 0.5) +  # Courbe de régression pour les deux continents
  # geom_point(data = predict_data, aes(y = predictedT, x = Temperature, color = "black"), size = 0.5) +
  labs(title = "Beta regression of inversion frequency on temperature",
       x = "Temperature (°C)", y = "Inversion frequency") +  # Titres
  scale_color_manual(values = c("E" = "steelblue3", "A" = "indianred3")) +  # Couleurs par continent
  scale_y_continuous(limits = c(0, 1)) +
  scale_x_continuous(limits = c(0, 12)) +
  theme_minimal()  # Thème minimal



plot_lat2 + plot_temp2