Veldproef_Script.R

library(ggplot2)
library(sf)
library(reshape2)
library(tidyverse)
library(devtools)
#devtools::install_github("inbo/fistools")
library(fistools)
library(googlesheets4)
library(camtraptor)
library(tidyverse)

#MOLENS GEYSKENS

#read data
getwd()
data<-read.csv("data/Data_Opname.csv",sep=";",header=T)
data$X<-NULL

data$Mass[which(data$Mass>70)]<-NA
data<-na.omit(data)


data$Date<-as.POSIXct(strptime(data$Date, format = "%d/%m/%Y"),tz="UTC")

ggplot(data,aes(x=Date,y=Fraction,group=as.factor(ID),colour=as.factor(ID)))+
  geom_line()+
  scale_color_discrete(name = "Baitbox")

ggplot(data,aes(x=Date,y=Mass,group=as.factor(ID),colour=as.factor(ID)))+
  geom_line(linewidth=1)+
  scale_color_discrete(name = "Baitbox")


# Natlandhoeve Ulbeek

data<-read.csv("data/Opname_Ulbeek.csv",sep=";",header=T)
colnames(data)<-c("Lokaaspunt","Moment0","Moment1","Moment2","Moment3","Moment4",
                  "Moment1","Moment2","Moment3","Moment4","Moment5","Moment6","Moment7",
                  "Moment1","Moment2")
data_pre<-data[,c(1,2:6)]
data_bestrijding<-data[,c(1,7:13)]
data_post<-data[,c(1,14:15)]

data_pre <- melt(data_pre, id.vars = "Lokaaspunt", variable.name = "Moment", value.name = "Gewicht")
data_pre$Week<-gsub("[^0-9]", "", data_pre$Moment)


data_bestrijding <- melt(data_bestrijding, id.vars = "Lokaaspunt", variable.name = "Moment", value.name = "Gewicht")
data_bestrijding$Dag<-gsub("[^0-9]", "", data_bestrijding$Moment)
start<-cbind.data.frame(seq(from=1,to=23,by=1),rep("a",length.out=23),rep(40,length.out=23),rep(0,length.out=23))
colnames(start)<-colnames(data_bestrijding)
data_bestrijding<-rbind.data.frame(data_bestrijding,start)

data_post <- melt(data_post, id.vars = "Lokaaspunt", variable.name = "Moment", value.name = "Gewicht")
data_post$Week<-gsub("[^0-9]", "", data_post$Moment)
start<-cbind.data.frame(seq(from=1,to=23,by=1),rep("a",length.out=23),rep(40,length.out=23),rep(0,length.out=23))
colnames(start)<-colnames(data_post)
data_post<-rbind.data.frame(data_post,start)

#Barplot bestrijding

summary_data <- data_bestrijding %>%
  group_by(Dag) %>%
  summarise(
    mean_gewicht = mean(Gewicht),
    sd_gewicht = sd(Gewicht),
    se_gewicht = sd(Gewicht) / sqrt(n())
  )

ggplot(summary_data, aes(x = Dag, y = mean_gewicht)) +
  geom_bar(stat = "identity", fill = "#F8766D") +
  geom_errorbar(aes(ymin = mean_gewicht - se_gewicht, ymax = mean_gewicht + se_gewicht), width = 0.2) +
  labs(title = "Bestrijding",
       x = "Dag",
       y = "Gemiddeld gewicht lokaas (g)") +
  theme_minimal()

#Barplot census

summary_data_pre <- na.omit(data_pre) %>%
  group_by(Week) %>%
  summarise(
    mean_gewicht = mean(Gewicht),
    sd_gewicht = sd(Gewicht),
    se_gewicht = sd(Gewicht) / sqrt(n())
  )

plot_pre<-ggplot(summary_data_pre, aes(x = Week, y = mean_gewicht)) +
  geom_bar(stat = "identity", fill = "#00BFC4") +
  geom_errorbar(aes(ymin = mean_gewicht - se_gewicht, ymax = mean_gewicht + se_gewicht), width = 0.2) +
  labs(title = "Pre-Census",
       x = "Week",
       y = "Gemiddeld gewicht lokaas (g)") +
  theme_minimal()

summary_data_post <- na.omit(data_post) %>%
  group_by(Week) %>%
  summarise(
    mean_gewicht = mean(Gewicht),
    sd_gewicht = sd(Gewicht),
    se_gewicht = sd(Gewicht) / sqrt(n())
  )

add<-summary_data_post[1:2,]
add$Week<-c(3,4)
add$mean_gewicht<-c(0,0)
add$sd_gewicht<-c(0,0)
add$se_gewicht<-c(0,0)
summary_data_post<-rbind.data.frame(summary_data_post,add)

plot_post<-ggplot(summary_data_post, aes(x = Week, y = mean_gewicht)) +
  geom_bar(stat = "identity", fill = "#00BFC4") +
  geom_errorbar(aes(ymin = mean_gewicht - se_gewicht, ymax = mean_gewicht + se_gewicht), width = 0.2) +
  labs(title = "Post-Census",
       x = "Week                            ",
       y = "") +
  theme_minimal()

library(gridExtra)
grid.arrange(plot_pre, plot_post, ncol = 2)

#Change from weight to opname

# Bestrijding
data_bestrijding$Opname<-40-data_bestrijding$Gewicht
data_bestrijding<-data_bestrijding[which(data_bestrijding$Dag!=0),]
summary_data <- data_bestrijding %>%
  group_by(Dag) %>%
  summarise(
    mean_opname = mean(Opname),
    sd_opname = sd(Opname),
    se_opname = sd(Opname) / sqrt(n())
  )

ggplot(summary_data, aes(x = Dag, y = mean_opname)) +
  geom_bar(stat = "identity", fill = "#F8766D") +
  geom_errorbar(aes(ymin = mean_opname - se_opname, ymax = mean_opname + se_opname), width = 0.2) +
  labs(title = "Bestrijding",
       x = "Dag",
       y = "Gemiddeld Opname lokaas(g)") +
  theme_minimal()+
  ylim(0,40)

#Pre census
data_pre$Opname<-40-data_pre$Gewicht
data_pre<-data_pre[which(data_pre$Week!=0),]
summary_data_pre <- na.omit(data_pre) %>%
  group_by(Week) %>%
  summarise(
    mean_Opname = mean(Opname),
    sd_Opname = sd(Opname),
    se_Opname = sd(Opname) / sqrt(n())
  )

plot_pre<-ggplot(summary_data_pre, aes(x = Week, y = mean_Opname)) +
  geom_bar(stat = "identity", fill = "#00BFC4") +
  geom_errorbar(aes(ymin = mean_Opname - se_Opname, ymax = mean_Opname + se_Opname), width = 0.2) +
  labs(title = "Pre-Census",
       x = "Week",
       y = "Gemiddeld Opname lokaas (g)") +
  theme_minimal()+
  ylim(0,40)

#Post
data_post$Opname<-40-data_post$Gewicht
data_post<-data_post[which(data_post$Week!=0),]
summary_data_post <- na.omit(data_post) %>%
  group_by(Week) %>%
  summarise(
    mean_Opname = mean(Opname),
    sd_Opname = sd(Opname),
    se_Opname = sd(Opname) / sqrt(n())
  )

add<-summary_data_post[1:2,]
add$Week<-c(3,4)
add$mean_Opname<-c(0,0)
add$sd_Opname<-c(0,0)
add$se_Opname<-c(0,0)
summary_data_post<-rbind.data.frame(summary_data_post,add)

plot_post<-ggplot(summary_data_post, aes(x = Week, y = mean_Opname)) +
  geom_bar(stat = "identity", fill = "orange") +
  geom_errorbar(aes(ymin = mean_Opname - se_Opname, ymax = mean_Opname + se_Opname), width = 0.2) +
  labs(title = "Post-Census",
       x = "Week                            ",
       y = "") +
  theme_minimal()+
  ylim(0,40)

library(gridExtra)
grid.arrange(plot_pre, plot_post, ncol = 2)

# Camera trap data analysis

# read data
readRenviron(".Renviron")
download_gdrive_if_missing(gfileID = "15nggsbct4WBW_57mw9bE69lfkntfciTV",
                           destfile = "data/veldproef-inbo-bruine-rat-20240813144457.zip",
                           email = Sys.getenv("EMAIL"),
                           update_always = TRUE)
unzip("data/veldproef-inbo-bruine-rat-20240813144457.zip",exdir="data/veldproef-inbo-bruine-rat-20240813144457/")
cam_data <- read_camtrap_dp(file="data/veldproef-inbo-bruine-rat-20240813144457/datapackage.json")

# Get  subsets
cam_precensus <- get_record_table(cam_data) %>% 
  filter(Species == "Rattus norvegicus",
         DateTimeOriginal < ymd_hms("2024-05-25 09:00:00")) %>% 
  mutate(test = "pre-census")

cam_postcensus <- get_record_table(cam_data) %>% 
  filter(Species == "Rattus norvegicus",
         DateTimeOriginal > ymd_hms("2024-05-30 09:00:00")) %>% 
  mutate(test = "post-census")

combined <- cam_precensus %>% 
  add_row(cam_postcensus)

ratten <- combined %>% 
  group_by(Station, test) %>% 
  summarise(n())

#Correct for duration of deployment (pre= 25/4 tot 23/5; post= 30/5 tot 13/6)
# 29 en 15 dagen
ratten$n_cor <- ifelse(ratten$test == "pre-census", ratten$`n()` / 29,
                       ifelse(ratten$test == "post-census", ratten$`n()` / 15, NA))

# Setting the factor levels for 'test' to ensure correct order
ratten$test <- factor(ratten$test, levels = c("pre-census", "post-census"))

#plot
ggplot(ratten, aes(x = Station, y = n_cor, fill = test)) +
  geom_bar(stat = "identity", position = position_dodge()) +
  labs(title = "",
       x = "Station",
       y = "Gemiddeld aantal bruine rat waarnemingen / dag",
       fill="Periode") +
  theme_minimal()

# Get  all data
cam <- get_record_table(cam_data) %>% 
  filter(Species == "Rattus norvegicus")

# summarize per day
summary_data <- cam %>%
  group_by(Station, Date) %>%
  summarize(total_observations = sum(n, na.rm = TRUE))

# add periods
summary_data$Period<-ifelse(as.POSIXct(summary_data$Date)<as.POSIXct("2024-05-25",format="%Y-%m-%d",tz="UTC"), "Pre-census",
                            ifelse(as.POSIXct(summary_data$Date)>=as.POSIXct("2024-05-30",format="%Y-%m-%d",tz="UTC"), "Post-census", NA))

#summarize
summary_data <- summary_data %>%
  group_by(Station, Period) %>%
  summarise(
    mean_observations = mean(total_observations),
    sem_observations = sd(total_observations) / sqrt(n())
  )

# Setting the factor levels for 'test' to ensure correct order
summary_data$Period <- factor(summary_data$Period, levels = c("Pre-census", "Post-census"))

#plot
ggplot(summary_data, aes(x = Station, y = mean_observations, fill = Period)) +
  geom_bar(stat = "identity", position = position_dodge(width = 0.9)) +
  geom_errorbar(aes(ymin = mean_observations - sem_observations, ymax = mean_observations + sem_observations),
                position = position_dodge(width = 0.9), width = 0.25) +
  labs(
    title = "",
    x = "Locatie",
    y = "Gemiddeld aantal waarnemingen bruine rat per dag",
    fill="Periode"
  ) +
  scale_fill_manual(values=c("#00BFC4","orange"))+
  theme_minimal()

# Add simulated 5 min interval between subsequent triggers

cam_interval <- cam %>%
  group_by(Station) %>%
  arrange(Station, DateTimeOriginal) %>%  # Make sure observations are ordered by time
  mutate(time_diff_secs = as.numeric(difftime(DateTimeOriginal, lag(DateTimeOriginal), units = "secs"))) %>%
  filter(is.na(time_diff_secs) | time_diff_secs >= 300)  # Keep first observation or those 5 minutes apart

# summarize per day
summary_data <- cam_interval %>%
  group_by(Station, Date) %>%
  summarize(total_observations = sum(n, na.rm = TRUE))

# add periods
summary_data$Period<-ifelse(as.POSIXct(summary_data$Date)<as.POSIXct("2024-05-25",format="%Y-%m-%d",tz="UTC"), "Pre-census",
                            ifelse(as.POSIXct(summary_data$Date)>=as.POSIXct("2024-05-30",format="%Y-%m-%d",tz="UTC"), "Post-census", NA))

#summarize
summary_data <- summary_data %>%
  group_by(Station, Period) %>%
  summarise(
    mean_observations = mean(total_observations),
    sem_observations = sd(total_observations) / sqrt(n())
  )

# Setting the factor levels for 'test' to ensure correct order
summary_data$Period <- factor(summary_data$Period, levels = c("Pre-census", "Post-census"))

#plot
plot_interval<-ggplot(summary_data, aes(x = Station, y = mean_observations, fill = Period)) +
  geom_bar(stat = "identity", position = position_dodge(width = 0.9)) +
  geom_errorbar(aes(ymin = mean_observations - sem_observations, ymax = mean_observations + sem_observations),
                position = position_dodge(width = 0.9), width = 0.25) +
  labs(
    title = "",
    x = "Locatie",
    y = "Gemiddeld aantal waarnemingen bruine rat per dag",
    fill="Periode"
  ) +
  scale_fill_manual(values=c("#00BFC4","orange"))+
  theme_minimal()


##################################################################################
####################  ANALYSE ####################################################
##################################################################################

library(glmmTMB)
library(DHARMa)

## Camera trap

# All data to check species
raw<-get_record_table(cam_data)
unique(raw$Species)

# Get rat data with 5 minute interval
cam_interval

# activity patterns

# Ensure DateTimeOriginal is in proper datetime format
cam_interval$DateTimeOriginal <- as.POSIXct(cam_interval$DateTimeOriginal)

# Extract the hour from DateTimeOriginal
cam_interval$hour <- format(cam_interval$DateTimeOriginal, "%H")

# Convert hour to numeric for plotting
cam_interval$hour <- as.numeric(cam_interval$hour)

# summarize per day
summary_data <- cam_interval %>%
  group_by(Station, Date) %>%
  summarize(total_observations = sum(n, na.rm = TRUE))

# add periods
summary_data$Period<-ifelse(as.POSIXct(summary_data$Date)<as.POSIXct("2024-05-25",format="%Y-%m-%d",tz="UTC"), "Pre-census",
                            ifelse(as.POSIXct(summary_data$Date)>=as.POSIXct("2024-05-30",format="%Y-%m-%d",tz="UTC"), "Post-census", NA))

# Calculate mean and standard deviation for each Period across all stations
summary_data %>%
  group_by(Period) %>%
  summarise(
    mean_observations = mean(total_observations),
    sd_observations = sd(total_observations)
  )

# Fit a Poisson GLMM
poisson_model <- glmmTMB(total_observations ~ Period + (1|Station), 
                       family = poisson, data = summary_data)

# Check overdispersion
overdispersion_value <- sum(residuals(poisson_model, type = "pearson")^2) / df.residual(poisson_model)
simulationOutput <- simulateResiduals(fittedModel = poisson_model, plot = F)
plot(simulationOutput)

# Use neg. binomial

model <- glmmTMB(total_observations ~ Period + (1|Station), 
                       family = nbinom1, data = summary_data)
simulationOutput <- simulateResiduals(fittedModel = model, plot = F)
plot(simulationOutput)

summary(model)


## Activiteit martijn 

library(activity)
library(overlap)

rec_table <-
  get_record_table(cam_data,
                   stationCol = "locationName",
                   exclude = NULL,
                   minDeltaTime = 0,
                   deltaTimeComparedTo = NULL
  )
rec_table<-
  subset(rec_table,!is.na(rec_table$Time))

rec_table<-rec_table%>% 
  filter(Species == "Rattus norvegicus")

lat<-rep(50.83624558822024,length=nrow(rec_table))
lon<-rep(5.287749867385671,length=nrow(rec_table))

coords<-matrix(c(lat,lon),ncol=2)

rec_table$date <-
  as.POSIXct(rec_table$Date, tz = "CET")

rec_table$clo <- gettime(rec_table$DateTimeOriginal)
rec_table$sunt <- sunTime(rec_table$clo,rec_table$DateTimeOriginal,coords)

ggplot(rec_table, aes(x = sunt)) +
  geom_density(size = 1.2) +
  labs(title = element_blank(),       # Remove title
       x = NULL,                      # Remove x-axis title
       y = "Waarnemingprobabiliteit",            # Keep y-axis label
       color = "Camera locatie") +     # Set the legend title for 'color'
  theme_minimal() +
  scale_x_continuous(breaks = c(0, pi/2, pi, 3*pi/2, 2*pi),  # Set custom axis ticks
                     labels = c("Middernacht", "Zonsopkomst", "Middag", "Zonsondergang", "Middernacht")) + 
  geom_vline(xintercept = pi/2, linetype = "dashed", color = "black", size = 1) +  # Zonsopkomst line
  geom_vline(xintercept = 3*pi/2, linetype = "dashed", color = "black", size = 1) +  # Zonsondergang line
  theme(legend.title = element_text(size = 12))

sunAct <- fitact(rec_table$sunt, reps = 200, sample = "data")

# sunAct$pdf contains the density data
pdf_data <- as.data.frame(sunAct@pdf)  # Convert the 'pdf' slot to a data frame
colnames(pdf_data) <- c("radians", "density", "se", "lcl", "ucl")  # Rename columns for clarity

# Create the plot
ggplot(pdf_data, aes(x = radians, y = density)) +
  geom_line(color = "black", size = 1.2) +  # Main density line
  geom_ribbon(aes(ymin = lcl, ymax = ucl), fill = "red", alpha = 0.2) +  # Confidence interval
  labs(title = element_blank(),       # Remove title
       x = NULL,
       y="Dichtheid")+                      # Remove x-axis title
  theme_minimal() +
  scale_x_continuous(breaks = c(0, pi/2, pi, 3*pi/2, 2*pi),  # Set custom axis ticks
                     labels = c("Middernacht", "Zonsopkomst", "Middag", "Zonsondergang", "Middernacht")) + 
  geom_vline(xintercept = pi/2, linetype = "dashed", color = "black", size = 1) +  # Zonsopkomst line
  geom_vline(xintercept = 3*pi/2, linetype = "dashed", color = "black", size = 1) +  # Zonsondergang line
  theme(legend.title = element_text(size = 12))


## Rodenticide

Weight<-c(data[,3],data[,4],data[,5],data[,6],data[,14],data[,15])
Opname<-40-Weight
Station<-rep(seq(1:23),6)
Period<-c(rep("Pre-census",length.out=nrow(data)*4),rep("Post-census",length.out=nrow(data)*2))
Bait<-cbind.data.frame(Station,Opname,Period)

# Make binary (eaten from bait =1, not eaten = 0)
Bait$Opname_Bin<-rep(1,length.out=nrow(Bait))
Bait$Opname_Bin[which(Bait$Opname == 0)]<- 0

bait_model <- glmmTMB(Opname_Bin ~ Period + (1|Station), 
                         family = binomial, data = Bait)
simulationOutput <- simulateResiduals(fittedModel = bait_model, plot = F)
plot(simulationOutput)
summary(bait_model)