The threat of wild boar in cultural sites is becoming a greater problem in recent years. The understanding of wild boar activity patterns is therefore of use. Using the data of wild boar movement from 15.12.2014 until the 15.01.2015 has been evaluated in the following manner: Different environments like (settlements, streets and different land use categories) have been contextualized and rankings have been derived answering questions like “What does a wild boar fear?” and “What are the behavioral constraints of wild boars?”. Further we answered questions like what is the most active, laziest and the most adventurous wild boar. Therefore we derived semi scientific psychological patterns on the individual level to conduct rankings based on the answered questions.
In recent decades, an expansion and increase of wild boar (Sus scrofa) populations has been observed across Europe, with colonizing agricultural and urban areas, thus increasing problems of cohabitation with humans (Rutten et al., 2020). In particular, the wild boar often comes into conflict with humans, the main problems being the transmission of diseases to domestic animals and humans, collisions with vehicles, disturbance or threat to citizens and damage to gardens, public parks, pasture and agriculture (Meng et al. 2009).
Wild boar damage to croplands is expressed in different ways: direct consumption of crops, rooting in search of bulbs, invertebrates or tubers, seed removal, trampling and damage to agricultural infrastructure (Barrios-García & Ballari 2012). Therefore, it is important to know their activity patterns and where and when the wild boars are in order to take countermeasures.
The research question asked is: How can the different environments (settlements, streets and different land use categories) being contextualized with the activity-patterns of the wild boars. How can this be conceptualized, modeled and further implemented in suitable visualizations to show behavioral patterns given to a certain environmental context and help to answer questions like “What does wild boar fear?” or “What are the behavioral constraints of wild boars?”. Further we evaluated questions like what is the most active, laziest and the most adventurous wild boar. Finally we derived semi scientific psychological patterns on the individual level to conduct rankings based on the answered questions.
For this project we needed the following data: wild boar data are given by the course administration of Patterns & Trends in Environmental Data, ZHAW, orthoimages & maps are given by swisstopo (open access) and the spatial data are given by geovite.ethz.ch (open access). An overview over the used data is given in table 1.
library(knitr)
<- data.frame(
data_aquisistion Data = c("Wild-Boar", "Orthophoto", "Map", "TLM_Landcover", "TLM_Streets", "TLM_Railway", "TLM_Building", "TLM_LeisureAreas"),
Data_Format = c(".csv", ".tif", ".tif", ".geojson", ".geojson", ".geojson", ".geojson", ".geojson"),
Geometry_Type = c("Points", "Raster", "Raster", "Polygon", "Lines", "Lines", "Polygon", "Polygon"),
Projected_CRS = c("CH1903+ / LV95", "CH1903 / LV03", "CH1903 / LV03", "WGS 84", "CH1903 / LV03", "CH1903 / LV03", "CH1903 / LV03", "CH1903 / LV03")
)
kable(data_aquisistion, col.names = gsub("[_]", " ", names(data_aquisistion)), caption = "Table 1: Overwiev of the used data for this project.")
Data | Data Format | Geometry Type | Projected CRS |
---|---|---|---|
Wild-Boar | .csv | Points | CH1903+ / LV95 |
Orthophoto | .tif | Raster | CH1903 / LV03 |
Map | .tif | Raster | CH1903 / LV03 |
TLM_Landcover | .geojson | Polygon | WGS 84 |
TLM_Streets | .geojson | Lines | CH1903 / LV03 |
TLM_Railway | .geojson | Lines | CH1903 / LV03 |
TLM_Building | .geojson | Polygon | CH1903 / LV03 |
TLM_LeisureAreas | .geojson | Polygon | CH1903 / LV03 |
After loading the necessary libraries and the data acquisition in R
(version 4.2.1), we converted their different coordinate reference
system into a consistent one (CH1903+ / LV95). Then, the data are
clipped to the study area with the extension of 2’567000/1’202000 to
2’573000/1’207000 with the function st_intersection()
. The
study area is located in the canton of Bern, district Seeland, Commune
Gempelen and is located near the lake Neuchâtel. To reduce the amount of
data volume - due to hardware limitation - only 7 of 19 wild boars were
taken and the time period from 15.12.2014 until the 15.01.2015 was
chosen. The result therefore is rather a simplified way to solve our
question but includes all the necessary data to answer our research
questions.
### Loading Libraries:
library(devtools)
library(ComputationalMovementAnalysisData)
library(ggplot2)
library(readr)
library(dplyr)
library(terra)
library(sf)
library(tmap)
library(SimilarityMeasures)
library(lubridate)
library(tidyr)
library(knitr)
library(zoo)
library(data.table)
### Data Import:
<- st_as_sf(wildschwein_BE, # Convert wild-boar csv-data in Spatial Object
Wildschwein_sf coords = c("E", "N"),
crs = 2056)
<- st_read("StudyArea.geojson") # Import Study Area
Study_Area <- rast("Swissimage_1m_2014.tif") # Import Orthophoto as Raster-Data
Orthophoto <- terra::rast("Map_1to25000_2018.tif") # Import Map as Raster-Data
Map <- st_read("Feldfruechte_1m_2021.geojson")
Feldfruechte <- st_read("TLM_Bodenbedeckung_2020.geojson") # Import Landcover-Types
TLM_Bodenbedeckung <- st_read("TLM_Strassen_2020.geojson") # Import Roads
TLM_Strassen <- st_read("TLM_Eisenbahn_2020.geojson") # Import Railway-Lines
TLM_Eisenbahn <- st_read("TLM_Gebaeude_Foodprint_2020.geojson") # Import Buildings
TLM_Gebaeude <- st_read("TLM_Freizeitareale_2020.geojson") # Import Leisure-Areas
TLM_Freizeitareale
### Convert CRS:
# Projected CRS: CH1903+ / LV95
Wildschwein_sf # Projected CRS: CH1903+ / LV95
Study_Area # Projected CRS: CH1903 / LV03
Orthophoto # Projected CRS: CH1903 / LV03
Map # Projected CRS: CH1903+ / LV95
Feldfruechte # Projected CRS: WGS 84
TLM_Bodenbedeckung # Projected CRS: CH1903 / LV03
TLM_Strassen # Projected CRS: CH1903 / LV03
TLM_Eisenbahn # Projected CRS: CH1903 / LV03
TLM_Gebaeude # Projected CRS: CH1903 / LV03
TLM_Freizeitareale
<- st_transform(TLM_Bodenbedeckung, crs = st_crs(2056))
TLM_Bodenbedeckung <- st_transform(TLM_Strassen, crs = st_crs(2056))
TLM_Strassen <- st_transform(TLM_Eisenbahn, crs = st_crs(2056))
TLM_Eisenbahn <- st_transform(TLM_Gebaeude, crs = st_crs(2056))
TLM_Gebaeude <- st_transform(TLM_Freizeitareale, crs = st_crs(2056))
TLM_Freizeitareale <- st_transform(Study_Area, crs = st_crs(21781))
Study_Area_for_Raster
### Objektart as Numeric:
$objektart <- as.numeric(TLM_Bodenbedeckung$objektart)
TLM_Bodenbedeckung$objektart <- as.numeric(TLM_Strassen$objektart)
TLM_Strassen$objektart <- as.numeric(TLM_Eisenbahn$objektart)
TLM_Eisenbahn$objektart <- as.numeric(TLM_Gebaeude$objektart)
TLM_Gebaeude$objektart <- as.numeric(TLM_Freizeitareale$objektart)
TLM_Freizeitareale
### Downsizing Orthophoto
<- aggregate(Orthophoto, fact=2)
Orthophoto
### Clip Data to Study_Area
<- st_intersection(Wildschwein_sf, Study_Area)
Wildschwein_sf <- crop(x = Orthophoto, y = Study_Area_for_Raster)
Orthophoto <- st_intersection(Feldfruechte, Study_Area)
Feldfruechte <- st_intersection(TLM_Bodenbedeckung, Study_Area)
TLM_Bodenbedeckung <- st_intersection(Study_Area, TLM_Strassen)
TLM_Strassen <- st_intersection(Study_Area, TLM_Eisenbahn)
TLM_Eisenbahn <- st_intersection(Study_Area, TLM_Gebaeude)
TLM_Gebaeude <- st_intersection(Study_Area, TLM_Freizeitareale)
TLM_Freizeitareale
### Choose Wild-Boar Data from 2014 to 2015
<- Wildschwein_sf %>%
Wildschwein_select filter(DatetimeUTC >= as.Date("2014-12-15") & DatetimeUTC < as.Date("2015-01-15"))
<- as.numeric(difftime(time1 = "2015-01-15", time2 = "2014-12-15", units = "secs"))
timespan
### Add Metadata
### Bodenbedeckung:
<- c(1:15)
objektart <- c("Fels", "Fels locker", "Felsbloecke", "Felsbloecke locker", "Fliessgewaesser", "Gebueschwald", "Lockergestein", "Lockergestein locker", "Gletscher", "Stehende Gewaesser", "Feuchtgebiet", "Wald", "Wald offen", "Gehoelzflaeche", "Schneefeld Toteis")
objektart_Beschreibung <- data.frame(objektart, objektart_Beschreibung)
Attributnamen
<-left_join(TLM_Bodenbedeckung, Attributnamen, by = "objektart")
TLM_Bodenbedeckung
### Strassen:
<- c(1:23)
objektart <- c("Ausfahrt", "Einfahrt", "Autobahn", "Raststaette", "Verbindung", "Zufahrt", "Dienstzufahrt", "10m Strasse", "6m Strasse", "4m Strasse", "3m Strasse", "Platz", "Autozug", "Faehre", "2m Weg", "1m Weg", "1m Wegfragment", "2m Wegfragment", "Markierte Spur","8m Strasse","Autostrasse","Klettersteig", "Provisorium")
objektart_Beschreibung <- data.frame(objektart, objektart_Beschreibung)
Attributnamen
<-left_join(TLM_Strassen, Attributnamen, by = "objektart")
TLM_Strassen
### Eisenbahn:
<- c(0,2,4,5)
objektart <- c("Normalspur", "Schmalspur", "Schmalspur mit Normalspur", "Kleinbahn")
objektart_Beschreibung <- data.frame(objektart, objektart_Beschreibung)
Attributnamen
<-left_join(TLM_Eisenbahn, Attributnamen, by = "objektart")
TLM_Eisenbahn
### Gebaeude:
<- c(0:22)
objektart <- c("Gebaeude", "NA", "Hochhaus", "Hochkamin", "Turm", "Kuehlturm", "Lagertank", "Lueftungsschacht", "Offenes Gebaeude", "Treibhaus", "Im Bau", "Kapelle", "Sakraler Turm", "Sakrales Gebaeude", "NA", "Flugdach", "Unterirdisches Gebaeude", "Mauer gross", "Mauer gross gedeckt", "Historische Baute", "NA", "NA", "Verbindungsbruecke")
objektart_Beschreibung <- data.frame(objektart, objektart_Beschreibung)
Attributnamen
<-left_join(TLM_Gebaeude, Attributnamen, by = "objektart")
TLM_Gebaeude
### Freizeitareale:
<- c(0:7)
objektart <- c("Campingplatzareal", "Freizeitanlagenareal", "Golfplatzareal", "Pferderennbahnareal", "Schwimmbadareal", "Sportplatzareal", "Standplatzareal", "Zooareal")
objektart_Beschreibung <- data.frame(objektart, objektart_Beschreibung)
Attributnamen
<-left_join(TLM_Freizeitareale, Attributnamen, by = "objektart") TLM_Freizeitareale
In order to find the most active, laziest or most adventurous individual, we derived the speed variable in order to find a proxy of the most active or laziest wild boar. Further we thought about another proxy of activeness: The net displacement to find the most adventurous individual over the entire study period (31 days). In this case to decided to derive the mean speed and mean displacement per one day intervals. In order to accomplish this task:
geom_tile()
.<- wildschwein_BE %>%
ws_newyear filter(DatetimeUTC >= as.Date("2014-12-15") & DatetimeUTC < as.Date("2015-01-15"))
<- ws_newyear %>%
ws_newyear mutate(timelag = as.numeric(difftime(lead(DatetimeUTC),DatetimeUTC,units = "mins")))
<- ws_newyear %>%
ws_newyear_steplength group_by(TierName) %>%
mutate(
steplength = sqrt((E-lead(E))^2+(N-lead(N))^2)
)
<- ws_newyear_steplength %>%
ws_newyear_speed group_by(TierName) %>%
mutate(
speed = steplength/timelag
)
<- ws_newyear %>%
ws_newyear_net_displacement mutate(
net_displacement = sqrt((E-lead(E, 20))^2+(N-lead(N, 20))^2)
)
%>%
ws_newyear_speed subset(select=-c(E,N)) %>%
mutate(datetime_round = lubridate::round_date(as.POSIXct(DatetimeUTC),unit = "1 day")) %>%
group_by(TierName, datetime_round) %>%
summarise(speed_mean = mean(speed)) -> ws_newyear_speed_mean
%>%
ws_newyear_steplength subset(select=-c(E,N)) %>%
mutate(datetime_round = lubridate::round_date(as.POSIXct(DatetimeUTC),unit = "1 day")) %>%
group_by(TierName, datetime_round) %>%
summarise(steplength_mean = mean(steplength)) -> ws_newyear_steplength_mean
%>%
ws_newyear_net_displacement subset(select=-c(E,N)) %>%
mutate(datetime_round = lubridate::round_date(as.POSIXct(DatetimeUTC),unit = "1 day")) %>%
group_by(TierName, datetime_round) %>%
summarise(net_displacement = mean(net_displacement)) -> ws_newyear_net_displacement
The idea behind the “Moving Window Approach” is to reduce the noise
(fluctuation) of the data. It therefore might be another approach to
analyze i.e. the mean speed variable but was not been further elaborated
in this project. The principal idea therefore is to sum up (in this case
six and five) variables like mean speed to smooth out movement peeks
(roll mean) over the given time period with the rollmean()
function from the zoo package. This allows to visualize the movement
patterns in a more trend driven way (Figure 1). Another idea would have
been to derive the roll-mean of the individuals and compare the overall
movement trends among each individual. This was not further been
elaborated because of the following more aesthetic and better working
approaches.
# Here we can see the overall movements of the seven wild boar in a smooth out way (derived by the rollmean function) and the original speedmean.
<- ws_newyear_speed_mean %>%
speed_mean head(100) %>%
mutate(rollmean = zoo::rollmean(speed_mean, 6, fill = NA, align = "center"))
<- melt(setDT(speed_mean), id.vars = c("TierName","datetime_round"), variable.name = c("Mean"))
speed_mean_long
<- ggplot(speed_mean_long, aes(x = datetime_round, y = value, color = Mean)) +
speed_mean_Plot geom_line() +
theme_classic() +
labs(y = "Speed [km/h]", x = "Time") +
guides(color = guide_legend(title = "Analyse Approach")) +
scale_color_discrete(labels = c("Mean Speed", "Moving Window Approach"))
speed_mean_Plot
# The individual approach in order to possibly compare them among each other. In this case all combined in one line-graph.
# i.e.:
<- ws_newyear_speed_mean %>%
individual head(37) %>%
mutate(rollmean = zoo::rollmean(speed_mean, 5, fill = NA, align = "center"))
<- ggplot(individual, aes(x = datetime_round, y = rollmean, color = "TierName")) +
individual_Plot geom_line() +
theme_classic() +
labs(y = "Roll Mean [km/h]", x = "Time")
individual_Plot
To describe the effect of the human influence areas at the activity-patterns of the wild boars, the question was asked: What is the most fearful and the most brave wild boar? The “fearful” wild boar is defined as the animal which avoids urban areas. The “brave” wild boar is the one that ventures closest to settlements. The ranking was based on the time spent in human influence areas as a percentage over the entire study period (31 days).
To define the human influence areas - such as streets, railway,
building and leisure areas - the data sets were first combined by
union()
. Then a buffer zone of 100 m was created around the
human influence areas (buffer()
). This range corresponds to
the escape distance of 100 m to 200 m of wild boars given in the
literature (Suter et al., 2018; Thurfjell et al. 2013). After joining
the wild boar data with the Human Influence Areas Buffers
(st_join()
), the time difference between subsequent rows
was calculated with difftime()
. All categories of the urban
areas were combined and named “Human Influence Areas” and all natural
environment as “Nature”. In Figure 2 a) are all human influence areas in
the study area shown and in b) the 100 m buffer zone around the human
influence areas.
### Combine Human Influence Areas.
### Geometry Type: Lines
<- TLM_Strassen[,c("objektart_Beschreibung", "geometry")]
Strassen_kl <- TLM_Eisenbahn[,c("objektart_Beschreibung", "geometry")]
Eisenbahn_kl
<- union(Strassen_kl, Eisenbahn_kl)
Lines <- st_as_sf(Lines)
Lines
### Geometry Type: Polygons
<- TLM_Gebaeude[,c("objektart_Beschreibung", "geometry")]
Gebaeude_kl <- TLM_Freizeitareale[,c("objektart_Beschreibung", "geometry")]
Freizeitareale_kl
<- union(Gebaeude_kl, Freizeitareale_kl)
Polygons <- st_as_sf(Polygons)
Polygons
### Total Human Influence Areas.
<- union(Lines, Polygons)
Human_Influence_Areas <- st_as_sf(Human_Influence_Areas)
Human_Influence_Areas
### Create Buffer around Human Influence Areas.
<- st_buffer(x = Human_Influence_Areas, dist = 100)
Human_Influence_Areas_Buffer
### Plot Human Influence Areas
<- tm_shape(shp = Orthophoto) +
Human_Influence_Areas_Plot tm_rgb() +
tm_shape(shp = Wildschwein_select) +
tm_dots(col = "TierName") +
tm_shape(shp = Human_Influence_Areas) +
tm_polygons(col = "objektart_Beschreibung") +
tm_lines(col = "objektart_Beschreibung") +
tm_compass(type = "arrow", position = c(0.247,0.028), bg.color = "white", bg.alpha = 0.75, size = 1.8) +
tm_scale_bar(breaks = c(0, 0.25, 0.5, 0.75 ,1), text.size = 0.75, position = c("left", "bottom"), bg.color = "white", bg.alpha = 0.75) +
tm_layout(main.title = "a)",
main.title.size = 1,
legend.title.size = 1,
legend.title.color = "white",
legend.text.size = 0.6,
legend.outside = TRUE,
legend.bg.color = "white",
legend.bg.alpha = 1)
<- mutate(Human_Influence_Areas_Buffer,Human_Influence_Areas = "Human Influence Areas")
Human_Influence_Areas_Buffer
<- tm_shape(shp = Orthophoto) +
Human_Influence_Areas_Buffer_Plot tm_rgb() +
tm_shape(shp = Human_Influence_Areas_Buffer) +
tm_polygons(col = "Human_Influence_Areas", palette = "grey") +
tm_shape(shp = Wildschwein_select) +
tm_dots(col = "TierName") +
tm_compass(type = "arrow", position = c(0.247,0.028), bg.color = "white", bg.alpha = 0.75, size = 1.8) +
tm_scale_bar(breaks = c(0, 0.25, 0.5, 0.75 ,1), text.size = 0.75, position = c("left", "bottom"), bg.color = "white", bg.alpha = 0.75) +
tm_layout(main.title = "b)",
main.title.size = 1,
legend.title.size = 1,
legend.title.color = "white",
legend.text.size = 0.6,
legend.outside = TRUE,
legend.bg.color = "white",
legend.bg.alpha = 1)
### most fearful/brave wild boar
<- st_join(x = Wildschwein_select, y = Human_Influence_Areas_Buffer)
Wildboar_brave
$objektart_Beschreibung <- replace_na(data = Wildboar_brave$objektart_Beschreibung, replace = "Wald&Wiesen&Natur")
Wildboar_brave
<- mutate(Wildboar_brave, Human_Influence_Areas = if_else(objektart_Beschreibung == "Wald&Wiesen&Natur", "Natural Habitats", "Human Influence Areas"))
Wildboar_brave
<- st_drop_geometry(Wildboar_brave)
Wildboar_brave
<- mutate(Wildboar_brave,time_of_stay = as.integer(difftime(time1 = lead(DatetimeUTC), time2 = DatetimeUTC), units = "secs"))
Wildboar_brave
<- Wildboar_brave[,c("TierName", "objektart_Beschreibung", "time_of_stay", "Human_Influence_Areas")]
Wildboar_brave
<- Wildboar_brave %>%
Wildboar_brave_list filter(time_of_stay >0) %>%
group_by(TierName, Human_Influence_Areas) %>%
summarise(time_of_stay_s = sum(time_of_stay)) %>%
mutate(time_of_stay_percent = (time_of_stay_s/timespan)*100)
Human_Influence_Areas_Plot
Human_Influence_Areas_Buffer_Plot
To describe the preferred locations - such as forest, water bodies,
wetlands and farmland - the wild boar data and the land cover data were
first joined by st_join()
. Then the time difference between
subsequent rows was calculated with difftime()
. The
classification was based on the time spent in the preferred locations as
a percentage over the entire study period (31 days). In Figure 3 are all
land cover types in the study area shown.
<- st_join(x = Wildschwein_select, y = TLM_Bodenbedeckung)
Wildboar_Bodenbedeckung
<- st_drop_geometry(Wildboar_Bodenbedeckung)
Wildboar_Bodenbedeckung
$objektart_Beschreibung <- replace_na(data = Wildboar_Bodenbedeckung$objektart_Beschreibung, replace = "Ackerland")
Wildboar_Bodenbedeckung
<- mutate(Wildboar_Bodenbedeckung,time_of_stay = as.integer(difftime(time1 = lead(DatetimeUTC), time2 = DatetimeUTC), units = "secs"))
Wildboar_Bodenbedeckung
<- Wildboar_Bodenbedeckung[,c("TierName", "objektart_Beschreibung", "time_of_stay")]
Wildboar_Bodenbedeckung
<- Wildboar_Bodenbedeckung %>%
Wildboar_Bodenbedeckung filter(time_of_stay >0) %>%
group_by(TierName, objektart_Beschreibung) %>%
summarise(time_of_stay_s = sum(time_of_stay)) %>%
mutate(time_of_stay_percent = (time_of_stay_s/timespan)*100)
# Plot
<- tm_shape(shp = Orthophoto) +
Wildboar_LandCover_Plot tm_rgb() +
tm_shape(shp = TLM_Bodenbedeckung) +
tm_polygons(col = "objektart_Beschreibung") +
tm_shape(shp = Wildschwein_select) +
tm_dots(col = "TierName") +
tm_compass(type = "arrow", position = c(0.247,0.028), bg.color = "white", bg.alpha = 0.75, size = 1.8) +
tm_scale_bar(breaks = c(0, 0.25, 0.5, 0.75 ,1), text.size = 0.75, position = c("left", "bottom"), bg.color = "white", bg.alpha = 0.75) +
tm_layout(main.title.size = 1,
legend.title.size = 1,
legend.title.color = "white",
legend.text.size = 0.6,
legend.outside = TRUE,
legend.bg.color = "white",
legend.bg.alpha = 1)
Wildboar_LandCover_Plot
The following are the core result of the question answered: which is the most active / laziest and the most adventurous wild boar. The results show clear “winners” and “losers” when having a look on the mean speed, mean step length and mean net displacement over the evaluated time period. We interpreted the results in the way, that the wild boar with the highest step length, speed or net displacement has been evaluated as the most active wild boar and vice versa. In order to support these findings, we also derived the overall mean speed and net displacement of each individual to back up our results and therefore finding the overall most active/laziest and the most adventurous wild boar.
# speed_mean
%>%
ws_newyear_speed_mean count(datetime_round, TierName) %>%
ggplot(mapping = aes(x = datetime_round, y = TierName)) +
geom_tile(mapping = aes(fill = ws_newyear_speed_mean$speed_mean)) +
theme_classic() +
labs(title = "a)", y = "Wild Boar Name", x = "Time") +
scale_fill_gradient(name = "Mean Speed [km/h]", low = "lightsalmon", high = "firebrick", na.value = "white")
# steplength_mean
%>%
ws_newyear_steplength_mean count(datetime_round, TierName) %>%
ggplot(mapping = aes(x = datetime_round, y = TierName)) +
geom_tile(mapping = aes(fill = ws_newyear_steplength_mean$steplength_mean)) +
theme_classic() +
labs(title = "b)", y = "Wild Boar Name", x = "Time") +
scale_fill_gradient(name = "Mean Step Length [m/day]", low = "lightsalmon", high = "firebrick", na.value = "white")
#net displacement
%>%
ws_newyear_net_displacement count(datetime_round, TierName) %>%
ggplot(mapping = aes(x = datetime_round, y = TierName)) +
geom_tile(mapping = aes(fill = ws_newyear_net_displacement$net_displacement)) +
theme_classic() +
labs(title = "c)", y = "Wild Boar Name", x = "Time") +
scale_fill_gradient(name = "Net Displacement [m/day]", low = "lightsalmon", high = "firebrick", na.value = "white")
The box plots in Figure 5 show the overall means of our indicators like speed/step length and net displacement of the individuals. According to the overall mean of net displacement, the winner seems to be the individual “Caroline.” However, the overall winner therefore is the individual “Fritz” whereas he has the highest mean speed, step length and 2end biggest net displacement over the hole time period. The overall most laziest wild boar therefore seems to be the individual Sabine. The exact values are given in Table 2.
ggplot(data = ws_newyear_speed_mean, aes(x=TierName, y = speed_mean)) +
geom_boxplot() +
theme_classic() +
labs(title = "a)", y = "Mean Speed [km/h]", x = "Time")
ggplot(data = ws_newyear_steplength_mean, aes(x=TierName, y = steplength_mean)) +
geom_boxplot() +
theme_classic() +
labs(title = "b)", y = "Mean Step Length [m/day]", x = "Time")
ggplot(data = ws_newyear_net_displacement, aes(x=TierName, y = net_displacement)) +
geom_boxplot() +
theme_classic() +
labs(title = "c)", y = "Net Displacement [m/day]", x = "Time")
<- ws_newyear_speed %>%
ranking_speed group_by(TierName) %>%
subset(select=-c(E,N)) %>%
summarise(Mean_Speed = round(mean(speed, na.rm =TRUE), digits = 2)) %>%
arrange(desc(Mean_Speed))
$Rank_speed <- 1:nrow(ranking_speed)
ranking_speed
<- ws_newyear_steplength %>%
ranking_steplength group_by(TierName) %>%
subset(select=-c(E,N)) %>%
summarise(Mean_Step_Length = round(mean(steplength, na.rm =TRUE), digits = 0)) %>%
arrange(desc(Mean_Step_Length))
$Rank_steplength <- 1:nrow(ranking_steplength)
ranking_steplength
<- ws_newyear_net_displacement %>%
ranking_displacement group_by(TierName) %>%
summarise(Net_Displacement = round(mean(net_displacement, na.rm =TRUE), digits = 0)) %>%
arrange(desc(Net_Displacement))
$Rank_displacement <- 1:nrow(ranking_displacement)
ranking_displacement
<- left_join(ranking_speed, ranking_steplength, by = "TierName")
R1
<- left_join(R1, ranking_displacement, by = "TierName")
R2
<- R2 %>%
R2 group_by(TierName, Mean_Speed, Mean_Step_Length, Net_Displacement) %>%
mutate(Rank = Rank_speed + Rank_steplength + Rank_displacement) %>%
arrange(Rank)
$Wild_Boar_Name <- R2$TierName
R2
<- R2[,c("Wild_Boar_Name", "Mean_Speed", "Mean_Step_Length", "Net_Displacement", "Rank")]
Ranking_active
$Rank <- 1:nrow(Ranking_active)
Ranking_active
kable(Ranking_active, col.names = gsub("[_]", " ", names(Ranking_active)), caption = "Table 2: The exact values and the ranking for the most active wild boar (Mean Speed [km/h], Mean Step Length [m/day], Net Displacement [m/day]) overall the study period (31 days).")
Wild Boar Name | Mean Speed | Mean Step Length | Net Displacement | Rank |
---|---|---|---|---|
Fritz | 2.21 | 160 | 453 | 1 |
Caroline | 1.64 | 114 | 478 | 2 |
Rosa | 1.61 | 80 | 333 | 3 |
Isabelle | 1.63 | 74 | 259 | 4 |
Nicole | 1.47 | 71 | 359 | 5 |
Ruth | 0.58 | 44 | 413 | 6 |
Sabine | 0.58 | 21 | 152 | 7 |
The effect of the human influence areas at the activity-patterns of the wild boars is shown in Figure 4. The duration of stay in urban areas is different for each animal. It is clearly seen, that the wild boars “Sabine” and “Ruth” venture closest to settlements. Whereas “Isabelle” and “Rosa” avoid the human influence areas and and prefer to stay in nature. The exact values and the ranking for each wild boar are given in Table 3 & 4.
# Plot
<- ggplot(Wildboar_brave_list, aes(x = TierName, y = time_of_stay_percent, fill = Human_Influence_Areas)) +
Wildboar_brave_list_plot geom_col() +
theme_classic() +
labs(y = "Duration of Stay [%]", x = "Wild Boar Name") +
guides(fill=guide_legend(title="Land Type")) +
scale_fill_manual(values = c("grey", "firebrick"))
Wildboar_brave_list_plot
# Ranking brave
<- Wildboar_brave_list %>%
Wildboar_brave_rank group_by(TierName) %>%
filter(Human_Influence_Areas == "Human Influence Areas") %>%
arrange(desc(time_of_stay_percent))
$Rank <- 1:nrow(Wildboar_brave_rank)
Wildboar_brave_rank$Wild_Boar_Name <- Wildboar_brave_rank$TierName
Wildboar_brave_rank$Duration_of_Stay <- Wildboar_brave_rank$time_of_stay_percent
Wildboar_brave_rank<- Wildboar_brave_rank[,c("Wild_Boar_Name", "Duration_of_Stay", "Rank")]
Wildboar_brave_rank $Duration_of_Stay <- round(Wildboar_brave_rank$Duration_of_Stay, digits = 0)
Wildboar_brave_rank
kable(Wildboar_brave_rank, col.names = gsub("[_]", " ", names(Wildboar_brave_rank)), caption = "Table 3: The exact values and the ranking for the brave wild boar (Duration of Stay in Percent).")
Wild Boar Name | Duration of Stay | Rank |
---|---|---|
Sabine | 98 | 1 |
Ruth | 97 | 2 |
Caroline | 75 | 3 |
Fritz | 53 | 4 |
Nicole | 27 | 5 |
Rosa | 14 | 6 |
Isabelle | 13 | 7 |
# Ranking fearful
<- Wildboar_brave_list %>%
Wildboar_fearful_rank group_by(TierName) %>%
filter(Human_Influence_Areas == "Natural Habitats") %>%
arrange(desc(time_of_stay_percent))
$Rank <- 1:nrow(Wildboar_fearful_rank)
Wildboar_fearful_rank$Wild_Boar_Name <- Wildboar_fearful_rank$TierName
Wildboar_fearful_rank$Duration_of_Stay <- Wildboar_fearful_rank$time_of_stay_percent
Wildboar_fearful_rank<- Wildboar_fearful_rank[,c("Wild_Boar_Name", "Duration_of_Stay", "Rank")]
Wildboar_fearful_rank $Duration_of_Stay <- round(Wildboar_fearful_rank$Duration_of_Stay, digits = 0)
Wildboar_fearful_rank
kable(Wildboar_fearful_rank, col.names = gsub("[_]", " ", names(Wildboar_fearful_rank)), caption = "Table 4: The exact values and the ranking for the fearful wild boar (Duration of Stay in Percent).")
Wild Boar Name | Duration of Stay | Rank |
---|---|---|
Isabelle | 87 | 1 |
Rosa | 86 | 2 |
Nicole | 73 | 3 |
Fritz | 47 | 4 |
Caroline | 25 | 5 |
Ruth | 3 | 6 |
Sabine | 2 | 7 |
The preferred locations of wild boars are very different and individual. “Sabine” and “Ruth” for example are water-loving, staying mostly in water bodies and wetlands. Whereas “Caroline” and “Fritz” are more in woods and farmlands. Over all, wetlands are the most popular location for every wild boar. The time stay in the land cover types is shown in Figure 5.
<- ggplot(Wildboar_Bodenbedeckung, aes(x = TierName, y = time_of_stay_percent, fill = objektart_Beschreibung)) +
Wildboar_LandType_Plot geom_col() +
theme_classic() +
labs(y = "Duration of Stay [%]", x = "Wild Boar Name") +
guides(fill=guide_legend(title="Land Type")) +
scale_fill_brewer(palette = 14)
Wildboar_LandType_Plot
Results and discussion of the research questions. Achievements and further steps.
In order to reduce the amount of data we chose 7 individuals over the time period of 15.12.2014 until the 15.01.2015. Our core results are the questions answered: Which is the most active / laziest and most adventurous wild boar. We therefore contextualized behavioral patterns given to a certain environment type (i.e. field, forest, settlement area) and answered the following questions: “What does a wild boar fear?” or “What are the behavioral constraints of wild boars?”.
The results showed clear “winners” and “losers” when having a look on the mean speed, mean step length and mean net displacement over the evaluated time period. We therefore interpreted the results in the way that the wild boar with the highest step length, speed got evaluated as the most active wild boar and vice versa. The wild boar with the highest net displacement got evaluated as the most adventurous.
The box plots show the overall mean proxies of our indicator like speed/step length and net displacement of the individuals. They seem to be the most adequate ways to visualize the overall activeness. The overall winner therefore seems to be the individual Fritz whereas he has the highest mean speed, step length and net displacement over the hole time period chosen. The overall laziest wild boar therefore seems to be the wild boar individual Sabine.
Interestingly whereas according to the overall mean of net displacement the winner seems to be the individual Caroline. We therefore added the title of the most adventurous wild boar equaling the most traveled wild boar to the individual Caroline.
The effect of the human influence areas at the activity-patterns of the wild boars is shown in Figure 2. The duration of close approximation to urban areas is different for each animal. The wild boars “Sabine” and “Ruth” venture closest to settlements. Whereas “Isabelle” and “Rosa” avoid close approximation to urban areas and prefer to stay in more remote areas like fields and forest. The exact values and the ranking for each wild boar are given in table 2 & 3.
Further steps for future research would be clearly to broaden the time period and therefore extending the number of individuals. The mean values could therefore be more informative because them bearing more overall information. Also, the contextualized movement patterns would expand the valuable number of psychological patterns because considering more overall data.
Report problems, limitations and solutions
The interpretation of what most “active” / “laziest” or “adventurous” means still remains a vague idea but in this project, we defined it as the fastest or most traveled wild boar over the given time period.
The Moving window approach in order to reduce noise of the data might have been another approach to analyze i.e. mean speed but was not further been elaborated in this project. The idea was to sum up – in this case six and five over the day mean variables like speed to smooth out movement peeks over the hole given time period of one month. This allows to visualize the movement patterns in a more trend driven way.
Data science choices
The derivation of speed, steplenght and net displacement has been done according to the theory covered in the lectures.
M. Noelia Barrios-Garcia, Sebastian A. Ballari, “Impact of wild boar (Sus scrofa) in its introduced and native range: a review”, Biol Invasions, vol, 14, pp. 2283–2300, 2012, https://link.springer.com/content/pdf/10.1007/s10530-012-0229-6.pdf.
X. J. Meng, D. S. Lindsay, and N. Sriranganathan, “Wild boars as sources for infectious diseases in livestock and humans”, Philosophical Transactions of the Royal Society B: Biological Sciences, vol. 364, no. 1530, pp. 2697–2707, 2009, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2865094/pdf/rstb20090086.pdf.
Anneleen Rutten, Jim Casaer, Diederik Strubbe and Herwig Leirs, “Agricultural and landscape factors related to increasing wild boar agricultural damage in a highly anthropogenic landscape”, Wildlife Biology, vol. 2020, no. 1, pp. 1-11, 2020, https://onlinelibrary.wiley.com/doi/pdf/10.2981/wlb.00634.
Henrik Thurfjell, Göran Spong & Göran Ericsson, “Effects of hunting on wild boar Sus scrofa behaviour”, Wildlife Biology, vol. 2020, no. 19, pp. 87-93, 2020, https://bioone.org/journals/wildlife-biology/volume-19/issue-1/12-027/Effects-of-hunting-on-wild-boar-Sus-scrofa-behaviour/10.2981/12-027.pdf.
Suter Stefan, Sandro Stoller & Benjamin Sigrist, “Prävention von Wildschweinschäden in der Landwirtschaft und Management von Wildschweinen in Schutzgebieten”, Schlussbericht, 2018, https://www.zhaw.ch/storage/hochschule/medien/bildmaterial/Schlussbericht_Wildschweinschreck.pdf.